diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4742 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 200,
+  "global_step": 668,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0029940119760479044,
+      "grad_norm": 2.2040421447560568,
+      "learning_rate": 9.999944704978835e-06,
+      "loss": 0.6026,
+      "step": 1
+    },
+    {
+      "epoch": 0.005988023952095809,
+      "grad_norm": 1.706829217782298,
+      "learning_rate": 9.999778821138357e-06,
+      "loss": 0.5602,
+      "step": 2
+    },
+    {
+      "epoch": 0.008982035928143712,
+      "grad_norm": 1.3201917385546982,
+      "learning_rate": 9.999502352147583e-06,
+      "loss": 0.5501,
+      "step": 3
+    },
+    {
+      "epoch": 0.011976047904191617,
+      "grad_norm": 1.4521425277487803,
+      "learning_rate": 9.999115304121459e-06,
+      "loss": 0.5781,
+      "step": 4
+    },
+    {
+      "epoch": 0.014970059880239521,
+      "grad_norm": 1.2061198483240583,
+      "learning_rate": 9.998617685620715e-06,
+      "loss": 0.4422,
+      "step": 5
+    },
+    {
+      "epoch": 0.017964071856287425,
+      "grad_norm": 1.2429550198024126,
+      "learning_rate": 9.998009507651683e-06,
+      "loss": 0.4971,
+      "step": 6
+    },
+    {
+      "epoch": 0.020958083832335328,
+      "grad_norm": 1.1286011870112085,
+      "learning_rate": 9.997290783666048e-06,
+      "loss": 0.4521,
+      "step": 7
+    },
+    {
+      "epoch": 0.023952095808383235,
+      "grad_norm": 1.1605139800048332,
+      "learning_rate": 9.996461529560553e-06,
+      "loss": 0.4348,
+      "step": 8
+    },
+    {
+      "epoch": 0.02694610778443114,
+      "grad_norm": 1.271946656514232,
+      "learning_rate": 9.995521763676645e-06,
+      "loss": 0.4915,
+      "step": 9
+    },
+    {
+      "epoch": 0.029940119760479042,
+      "grad_norm": 1.2493769539888613,
+      "learning_rate": 9.994471506800078e-06,
+      "loss": 0.4464,
+      "step": 10
+    },
+    {
+      "epoch": 0.03293413173652695,
+      "grad_norm": 1.064126963343827,
+      "learning_rate": 9.993310782160439e-06,
+      "loss": 0.4809,
+      "step": 11
+    },
+    {
+      "epoch": 0.03592814371257485,
+      "grad_norm": 0.9959531195978738,
+      "learning_rate": 9.992039615430648e-06,
+      "loss": 0.4447,
+      "step": 12
+    },
+    {
+      "epoch": 0.038922155688622756,
+      "grad_norm": 0.9543314290660774,
+      "learning_rate": 9.99065803472638e-06,
+      "loss": 0.4216,
+      "step": 13
+    },
+    {
+      "epoch": 0.041916167664670656,
+      "grad_norm": 0.9850485715677441,
+      "learning_rate": 9.989166070605447e-06,
+      "loss": 0.4095,
+      "step": 14
+    },
+    {
+      "epoch": 0.04491017964071856,
+      "grad_norm": 0.9009609876478459,
+      "learning_rate": 9.98756375606713e-06,
+      "loss": 0.395,
+      "step": 15
+    },
+    {
+      "epoch": 0.04790419161676647,
+      "grad_norm": 1.0123732065428233,
+      "learning_rate": 9.985851126551428e-06,
+      "loss": 0.4286,
+      "step": 16
+    },
+    {
+      "epoch": 0.05089820359281437,
+      "grad_norm": 0.7433848076880695,
+      "learning_rate": 9.9840282199383e-06,
+      "loss": 0.3522,
+      "step": 17
+    },
+    {
+      "epoch": 0.05389221556886228,
+      "grad_norm": 0.7868585864748767,
+      "learning_rate": 9.982095076546806e-06,
+      "loss": 0.3991,
+      "step": 18
+    },
+    {
+      "epoch": 0.05688622754491018,
+      "grad_norm": 0.9349585272092751,
+      "learning_rate": 9.980051739134235e-06,
+      "loss": 0.4102,
+      "step": 19
+    },
+    {
+      "epoch": 0.059880239520958084,
+      "grad_norm": 0.9536042984951868,
+      "learning_rate": 9.977898252895133e-06,
+      "loss": 0.4178,
+      "step": 20
+    },
+    {
+      "epoch": 0.06287425149700598,
+      "grad_norm": 0.8474430314327542,
+      "learning_rate": 9.975634665460333e-06,
+      "loss": 0.3971,
+      "step": 21
+    },
+    {
+      "epoch": 0.0658682634730539,
+      "grad_norm": 0.9216923558311936,
+      "learning_rate": 9.973261026895878e-06,
+      "loss": 0.3639,
+      "step": 22
+    },
+    {
+      "epoch": 0.0688622754491018,
+      "grad_norm": 0.8986215167386543,
+      "learning_rate": 9.970777389701927e-06,
+      "loss": 0.449,
+      "step": 23
+    },
+    {
+      "epoch": 0.0718562874251497,
+      "grad_norm": 0.8226404900460093,
+      "learning_rate": 9.968183808811586e-06,
+      "loss": 0.3538,
+      "step": 24
+    },
+    {
+      "epoch": 0.0748502994011976,
+      "grad_norm": 0.9630832199680331,
+      "learning_rate": 9.965480341589702e-06,
+      "loss": 0.4195,
+      "step": 25
+    },
+    {
+      "epoch": 0.07784431137724551,
+      "grad_norm": 0.826812637560728,
+      "learning_rate": 9.962667047831585e-06,
+      "loss": 0.4146,
+      "step": 26
+    },
+    {
+      "epoch": 0.08083832335329341,
+      "grad_norm": 0.8668455064895977,
+      "learning_rate": 9.95974398976169e-06,
+      "loss": 0.3661,
+      "step": 27
+    },
+    {
+      "epoch": 0.08383233532934131,
+      "grad_norm": 0.9200527883911036,
+      "learning_rate": 9.95671123203224e-06,
+      "loss": 0.4069,
+      "step": 28
+    },
+    {
+      "epoch": 0.08682634730538923,
+      "grad_norm": 0.9692196219649082,
+      "learning_rate": 9.953568841721796e-06,
+      "loss": 0.4349,
+      "step": 29
+    },
+    {
+      "epoch": 0.08982035928143713,
+      "grad_norm": 0.775987435979951,
+      "learning_rate": 9.950316888333775e-06,
+      "loss": 0.3542,
+      "step": 30
+    },
+    {
+      "epoch": 0.09281437125748503,
+      "grad_norm": 0.8178557241224351,
+      "learning_rate": 9.946955443794908e-06,
+      "loss": 0.4216,
+      "step": 31
+    },
+    {
+      "epoch": 0.09580838323353294,
+      "grad_norm": 0.9389607587057774,
+      "learning_rate": 9.943484582453653e-06,
+      "loss": 0.4044,
+      "step": 32
+    },
+    {
+      "epoch": 0.09880239520958084,
+      "grad_norm": 0.7794272282393406,
+      "learning_rate": 9.939904381078553e-06,
+      "loss": 0.3491,
+      "step": 33
+    },
+    {
+      "epoch": 0.10179640718562874,
+      "grad_norm": 0.7020980849021461,
+      "learning_rate": 9.93621491885653e-06,
+      "loss": 0.3271,
+      "step": 34
+    },
+    {
+      "epoch": 0.10479041916167664,
+      "grad_norm": 0.9131827948904151,
+      "learning_rate": 9.932416277391144e-06,
+      "loss": 0.3765,
+      "step": 35
+    },
+    {
+      "epoch": 0.10778443113772455,
+      "grad_norm": 0.8839063900324204,
+      "learning_rate": 9.928508540700775e-06,
+      "loss": 0.3889,
+      "step": 36
+    },
+    {
+      "epoch": 0.11077844311377245,
+      "grad_norm": 0.864348313953269,
+      "learning_rate": 9.924491795216777e-06,
+      "loss": 0.4243,
+      "step": 37
+    },
+    {
+      "epoch": 0.11377245508982035,
+      "grad_norm": 0.8862059070465331,
+      "learning_rate": 9.920366129781564e-06,
+      "loss": 0.3956,
+      "step": 38
+    },
+    {
+      "epoch": 0.11676646706586827,
+      "grad_norm": 0.7131817643346815,
+      "learning_rate": 9.916131635646635e-06,
+      "loss": 0.3475,
+      "step": 39
+    },
+    {
+      "epoch": 0.11976047904191617,
+      "grad_norm": 0.8942423501970304,
+      "learning_rate": 9.91178840647057e-06,
+      "loss": 0.4226,
+      "step": 40
+    },
+    {
+      "epoch": 0.12275449101796407,
+      "grad_norm": 0.8130987873905524,
+      "learning_rate": 9.907336538316946e-06,
+      "loss": 0.3674,
+      "step": 41
+    },
+    {
+      "epoch": 0.12574850299401197,
+      "grad_norm": 0.7581674275510644,
+      "learning_rate": 9.902776129652223e-06,
+      "loss": 0.3211,
+      "step": 42
+    },
+    {
+      "epoch": 0.12874251497005987,
+      "grad_norm": 0.835347490776918,
+      "learning_rate": 9.898107281343557e-06,
+      "loss": 0.3332,
+      "step": 43
+    },
+    {
+      "epoch": 0.1317365269461078,
+      "grad_norm": 0.7603515058761441,
+      "learning_rate": 9.893330096656576e-06,
+      "loss": 0.3115,
+      "step": 44
+    },
+    {
+      "epoch": 0.1347305389221557,
+      "grad_norm": 0.8306219963480352,
+      "learning_rate": 9.888444681253087e-06,
+      "loss": 0.3832,
+      "step": 45
+    },
+    {
+      "epoch": 0.1377245508982036,
+      "grad_norm": 0.8958770540240665,
+      "learning_rate": 9.883451143188753e-06,
+      "loss": 0.3802,
+      "step": 46
+    },
+    {
+      "epoch": 0.1407185628742515,
+      "grad_norm": 0.8066887289512282,
+      "learning_rate": 9.878349592910694e-06,
+      "loss": 0.3172,
+      "step": 47
+    },
+    {
+      "epoch": 0.1437125748502994,
+      "grad_norm": 0.8836303821107435,
+      "learning_rate": 9.873140143255035e-06,
+      "loss": 0.406,
+      "step": 48
+    },
+    {
+      "epoch": 0.1467065868263473,
+      "grad_norm": 0.8297081568246096,
+      "learning_rate": 9.867822909444435e-06,
+      "loss": 0.3981,
+      "step": 49
+    },
+    {
+      "epoch": 0.1497005988023952,
+      "grad_norm": 0.845789689226809,
+      "learning_rate": 9.862398009085511e-06,
+      "loss": 0.3701,
+      "step": 50
+    },
+    {
+      "epoch": 0.15269461077844312,
+      "grad_norm": 0.818065405374004,
+      "learning_rate": 9.856865562166256e-06,
+      "loss": 0.3493,
+      "step": 51
+    },
+    {
+      "epoch": 0.15568862275449102,
+      "grad_norm": 0.7540749538935108,
+      "learning_rate": 9.851225691053382e-06,
+      "loss": 0.3184,
+      "step": 52
+    },
+    {
+      "epoch": 0.15868263473053892,
+      "grad_norm": 0.8856186820008544,
+      "learning_rate": 9.8454785204896e-06,
+      "loss": 0.3891,
+      "step": 53
+    },
+    {
+      "epoch": 0.16167664670658682,
+      "grad_norm": 0.7122037264181366,
+      "learning_rate": 9.83962417759088e-06,
+      "loss": 0.3317,
+      "step": 54
+    },
+    {
+      "epoch": 0.16467065868263472,
+      "grad_norm": 0.9025984473960144,
+      "learning_rate": 9.833662791843628e-06,
+      "loss": 0.334,
+      "step": 55
+    },
+    {
+      "epoch": 0.16766467065868262,
+      "grad_norm": 0.8296926651602788,
+      "learning_rate": 9.827594495101824e-06,
+      "loss": 0.3502,
+      "step": 56
+    },
+    {
+      "epoch": 0.17065868263473055,
+      "grad_norm": 0.8368614929206053,
+      "learning_rate": 9.821419421584108e-06,
+      "loss": 0.3657,
+      "step": 57
+    },
+    {
+      "epoch": 0.17365269461077845,
+      "grad_norm": 0.8024299882215358,
+      "learning_rate": 9.815137707870806e-06,
+      "loss": 0.3932,
+      "step": 58
+    },
+    {
+      "epoch": 0.17664670658682635,
+      "grad_norm": 0.9355249953810083,
+      "learning_rate": 9.808749492900917e-06,
+      "loss": 0.3546,
+      "step": 59
+    },
+    {
+      "epoch": 0.17964071856287425,
+      "grad_norm": 0.7741694345084122,
+      "learning_rate": 9.802254917969033e-06,
+      "loss": 0.3698,
+      "step": 60
+    },
+    {
+      "epoch": 0.18263473053892215,
+      "grad_norm": 0.909380338511485,
+      "learning_rate": 9.795654126722218e-06,
+      "loss": 0.362,
+      "step": 61
+    },
+    {
+      "epoch": 0.18562874251497005,
+      "grad_norm": 0.7781447173070738,
+      "learning_rate": 9.788947265156828e-06,
+      "loss": 0.3441,
+      "step": 62
+    },
+    {
+      "epoch": 0.18862275449101795,
+      "grad_norm": 0.8417092598843253,
+      "learning_rate": 9.782134481615282e-06,
+      "loss": 0.3191,
+      "step": 63
+    },
+    {
+      "epoch": 0.19161676646706588,
+      "grad_norm": 0.824159668729467,
+      "learning_rate": 9.775215926782788e-06,
+      "loss": 0.3858,
+      "step": 64
+    },
+    {
+      "epoch": 0.19461077844311378,
+      "grad_norm": 0.8778743074955353,
+      "learning_rate": 9.768191753683997e-06,
+      "loss": 0.3467,
+      "step": 65
+    },
+    {
+      "epoch": 0.19760479041916168,
+      "grad_norm": 0.8724861938818232,
+      "learning_rate": 9.761062117679632e-06,
+      "loss": 0.3438,
+      "step": 66
+    },
+    {
+      "epoch": 0.20059880239520958,
+      "grad_norm": 0.8197526384638352,
+      "learning_rate": 9.75382717646304e-06,
+      "loss": 0.3246,
+      "step": 67
+    },
+    {
+      "epoch": 0.20359281437125748,
+      "grad_norm": 0.9092970904697121,
+      "learning_rate": 9.746487090056712e-06,
+      "loss": 0.4084,
+      "step": 68
+    },
+    {
+      "epoch": 0.20658682634730538,
+      "grad_norm": 0.7222055598812208,
+      "learning_rate": 9.739042020808746e-06,
+      "loss": 0.3102,
+      "step": 69
+    },
+    {
+      "epoch": 0.20958083832335328,
+      "grad_norm": 0.8524180894036101,
+      "learning_rate": 9.73149213338924e-06,
+      "loss": 0.3904,
+      "step": 70
+    },
+    {
+      "epoch": 0.2125748502994012,
+      "grad_norm": 0.8555322805223762,
+      "learning_rate": 9.72383759478667e-06,
+      "loss": 0.3869,
+      "step": 71
+    },
+    {
+      "epoch": 0.2155688622754491,
+      "grad_norm": 0.8536310984136708,
+      "learning_rate": 9.71607857430419e-06,
+      "loss": 0.3024,
+      "step": 72
+    },
+    {
+      "epoch": 0.218562874251497,
+      "grad_norm": 0.8936249808386063,
+      "learning_rate": 9.708215243555875e-06,
+      "loss": 0.379,
+      "step": 73
+    },
+    {
+      "epoch": 0.2215568862275449,
+      "grad_norm": 0.9789467528433705,
+      "learning_rate": 9.700247776462944e-06,
+      "loss": 0.3625,
+      "step": 74
+    },
+    {
+      "epoch": 0.2245508982035928,
+      "grad_norm": 0.9208323679997393,
+      "learning_rate": 9.6921763492499e-06,
+      "loss": 0.3526,
+      "step": 75
+    },
+    {
+      "epoch": 0.2275449101796407,
+      "grad_norm": 0.7732918942210613,
+      "learning_rate": 9.68400114044064e-06,
+      "loss": 0.2822,
+      "step": 76
+    },
+    {
+      "epoch": 0.23053892215568864,
+      "grad_norm": 0.8819282042071043,
+      "learning_rate": 9.6757223308545e-06,
+      "loss": 0.3537,
+      "step": 77
+    },
+    {
+      "epoch": 0.23353293413173654,
+      "grad_norm": 0.8614139255257659,
+      "learning_rate": 9.667340103602263e-06,
+      "loss": 0.3994,
+      "step": 78
+    },
+    {
+      "epoch": 0.23652694610778444,
+      "grad_norm": 0.7935921391382667,
+      "learning_rate": 9.658854644082099e-06,
+      "loss": 0.3196,
+      "step": 79
+    },
+    {
+      "epoch": 0.23952095808383234,
+      "grad_norm": 1.0594181520845294,
+      "learning_rate": 9.650266139975474e-06,
+      "loss": 0.3903,
+      "step": 80
+    },
+    {
+      "epoch": 0.24251497005988024,
+      "grad_norm": 0.9369633580310035,
+      "learning_rate": 9.641574781242999e-06,
+      "loss": 0.3442,
+      "step": 81
+    },
+    {
+      "epoch": 0.24550898203592814,
+      "grad_norm": 0.8687848684448142,
+      "learning_rate": 9.632780760120217e-06,
+      "loss": 0.3542,
+      "step": 82
+    },
+    {
+      "epoch": 0.24850299401197604,
+      "grad_norm": 0.8617960569259269,
+      "learning_rate": 9.62388427111336e-06,
+      "loss": 0.3313,
+      "step": 83
+    },
+    {
+      "epoch": 0.25149700598802394,
+      "grad_norm": 0.7778268159117772,
+      "learning_rate": 9.614885510995047e-06,
+      "loss": 0.2961,
+      "step": 84
+    },
+    {
+      "epoch": 0.25449101796407186,
+      "grad_norm": 0.8919894825077804,
+      "learning_rate": 9.605784678799934e-06,
+      "loss": 0.3389,
+      "step": 85
+    },
+    {
+      "epoch": 0.25748502994011974,
+      "grad_norm": 0.9124988709452522,
+      "learning_rate": 9.596581975820304e-06,
+      "loss": 0.3487,
+      "step": 86
+    },
+    {
+      "epoch": 0.26047904191616766,
+      "grad_norm": 0.9751886763505696,
+      "learning_rate": 9.587277605601617e-06,
+      "loss": 0.3731,
+      "step": 87
+    },
+    {
+      "epoch": 0.2634730538922156,
+      "grad_norm": 0.960018724281691,
+      "learning_rate": 9.577871773938013e-06,
+      "loss": 0.3887,
+      "step": 88
+    },
+    {
+      "epoch": 0.26646706586826346,
+      "grad_norm": 0.9001216026084041,
+      "learning_rate": 9.568364688867757e-06,
+      "loss": 0.4347,
+      "step": 89
+    },
+    {
+      "epoch": 0.2694610778443114,
+      "grad_norm": 0.8628325506806731,
+      "learning_rate": 9.558756560668637e-06,
+      "loss": 0.3325,
+      "step": 90
+    },
+    {
+      "epoch": 0.27245508982035926,
+      "grad_norm": 0.9403070905201559,
+      "learning_rate": 9.549047601853313e-06,
+      "loss": 0.3424,
+      "step": 91
+    },
+    {
+      "epoch": 0.2754491017964072,
+      "grad_norm": 0.9757818110255363,
+      "learning_rate": 9.539238027164618e-06,
+      "loss": 0.3676,
+      "step": 92
+    },
+    {
+      "epoch": 0.27844311377245506,
+      "grad_norm": 0.8919717072992648,
+      "learning_rate": 9.52932805357081e-06,
+      "loss": 0.3521,
+      "step": 93
+    },
+    {
+      "epoch": 0.281437125748503,
+      "grad_norm": 0.8693087618050774,
+      "learning_rate": 9.519317900260769e-06,
+      "loss": 0.3806,
+      "step": 94
+    },
+    {
+      "epoch": 0.2844311377245509,
+      "grad_norm": 0.8634269398769101,
+      "learning_rate": 9.509207788639148e-06,
+      "loss": 0.3694,
+      "step": 95
+    },
+    {
+      "epoch": 0.2874251497005988,
+      "grad_norm": 0.8463285500330672,
+      "learning_rate": 9.498997942321484e-06,
+      "loss": 0.367,
+      "step": 96
+    },
+    {
+      "epoch": 0.2904191616766467,
+      "grad_norm": 0.7635633808551265,
+      "learning_rate": 9.488688587129243e-06,
+      "loss": 0.3852,
+      "step": 97
+    },
+    {
+      "epoch": 0.2934131736526946,
+      "grad_norm": 0.9900918694741316,
+      "learning_rate": 9.47827995108483e-06,
+      "loss": 0.3355,
+      "step": 98
+    },
+    {
+      "epoch": 0.2964071856287425,
+      "grad_norm": 0.996999918546389,
+      "learning_rate": 9.467772264406545e-06,
+      "loss": 0.3528,
+      "step": 99
+    },
+    {
+      "epoch": 0.2994011976047904,
+      "grad_norm": 0.7864973715714452,
+      "learning_rate": 9.457165759503492e-06,
+      "loss": 0.3191,
+      "step": 100
+    },
+    {
+      "epoch": 0.3023952095808383,
+      "grad_norm": 0.8771419186101935,
+      "learning_rate": 9.446460670970436e-06,
+      "loss": 0.3286,
+      "step": 101
+    },
+    {
+      "epoch": 0.30538922155688625,
+      "grad_norm": 0.866728406911566,
+      "learning_rate": 9.435657235582616e-06,
+      "loss": 0.2821,
+      "step": 102
+    },
+    {
+      "epoch": 0.3083832335329341,
+      "grad_norm": 0.9688078291261988,
+      "learning_rate": 9.424755692290507e-06,
+      "loss": 0.3543,
+      "step": 103
+    },
+    {
+      "epoch": 0.31137724550898205,
+      "grad_norm": 0.9286852806387357,
+      "learning_rate": 9.413756282214538e-06,
+      "loss": 0.3818,
+      "step": 104
+    },
+    {
+      "epoch": 0.3143712574850299,
+      "grad_norm": 0.8009276506295196,
+      "learning_rate": 9.402659248639749e-06,
+      "loss": 0.3415,
+      "step": 105
+    },
+    {
+      "epoch": 0.31736526946107785,
+      "grad_norm": 0.8759097193857428,
+      "learning_rate": 9.391464837010428e-06,
+      "loss": 0.3349,
+      "step": 106
+    },
+    {
+      "epoch": 0.3203592814371258,
+      "grad_norm": 0.9227104063780066,
+      "learning_rate": 9.380173294924661e-06,
+      "loss": 0.3866,
+      "step": 107
+    },
+    {
+      "epoch": 0.32335329341317365,
+      "grad_norm": 0.8649217141295237,
+      "learning_rate": 9.368784872128877e-06,
+      "loss": 0.3449,
+      "step": 108
+    },
+    {
+      "epoch": 0.3263473053892216,
+      "grad_norm": 0.8417111018584573,
+      "learning_rate": 9.357299820512305e-06,
+      "loss": 0.3312,
+      "step": 109
+    },
+    {
+      "epoch": 0.32934131736526945,
+      "grad_norm": 0.8645182101127776,
+      "learning_rate": 9.345718394101412e-06,
+      "loss": 0.3341,
+      "step": 110
+    },
+    {
+      "epoch": 0.3323353293413174,
+      "grad_norm": 0.8915971937099609,
+      "learning_rate": 9.334040849054288e-06,
+      "loss": 0.3727,
+      "step": 111
+    },
+    {
+      "epoch": 0.33532934131736525,
+      "grad_norm": 0.8962847413894931,
+      "learning_rate": 9.322267443654974e-06,
+      "loss": 0.3439,
+      "step": 112
+    },
+    {
+      "epoch": 0.3383233532934132,
+      "grad_norm": 0.8992657693893839,
+      "learning_rate": 9.310398438307747e-06,
+      "loss": 0.3907,
+      "step": 113
+    },
+    {
+      "epoch": 0.3413173652694611,
+      "grad_norm": 0.8250515120084873,
+      "learning_rate": 9.29843409553137e-06,
+      "loss": 0.3188,
+      "step": 114
+    },
+    {
+      "epoch": 0.344311377245509,
+      "grad_norm": 0.7732709676855889,
+      "learning_rate": 9.286374679953278e-06,
+      "loss": 0.3221,
+      "step": 115
+    },
+    {
+      "epoch": 0.3473053892215569,
+      "grad_norm": 0.8340614618224027,
+      "learning_rate": 9.274220458303727e-06,
+      "loss": 0.3033,
+      "step": 116
+    },
+    {
+      "epoch": 0.3502994011976048,
+      "grad_norm": 0.9081519043317762,
+      "learning_rate": 9.261971699409893e-06,
+      "loss": 0.3596,
+      "step": 117
+    },
+    {
+      "epoch": 0.3532934131736527,
+      "grad_norm": 0.9804451115483439,
+      "learning_rate": 9.249628674189928e-06,
+      "loss": 0.3401,
+      "step": 118
+    },
+    {
+      "epoch": 0.3562874251497006,
+      "grad_norm": 0.9048286976335761,
+      "learning_rate": 9.237191655646972e-06,
+      "loss": 0.3683,
+      "step": 119
+    },
+    {
+      "epoch": 0.3592814371257485,
+      "grad_norm": 0.9481229645975885,
+      "learning_rate": 9.224660918863104e-06,
+      "loss": 0.3976,
+      "step": 120
+    },
+    {
+      "epoch": 0.36227544910179643,
+      "grad_norm": 0.9306034331841647,
+      "learning_rate": 9.212036740993265e-06,
+      "loss": 0.3466,
+      "step": 121
+    },
+    {
+      "epoch": 0.3652694610778443,
+      "grad_norm": 0.9448818501487511,
+      "learning_rate": 9.199319401259132e-06,
+      "loss": 0.3531,
+      "step": 122
+    },
+    {
+      "epoch": 0.36826347305389223,
+      "grad_norm": 0.9092421622239264,
+      "learning_rate": 9.186509180942928e-06,
+      "loss": 0.3278,
+      "step": 123
+    },
+    {
+      "epoch": 0.3712574850299401,
+      "grad_norm": 0.8789641383722142,
+      "learning_rate": 9.173606363381218e-06,
+      "loss": 0.39,
+      "step": 124
+    },
+    {
+      "epoch": 0.37425149700598803,
+      "grad_norm": 0.9024575753731743,
+      "learning_rate": 9.16061123395863e-06,
+      "loss": 0.3488,
+      "step": 125
+    },
+    {
+      "epoch": 0.3772455089820359,
+      "grad_norm": 0.8869529616369447,
+      "learning_rate": 9.147524080101543e-06,
+      "loss": 0.3438,
+      "step": 126
+    },
+    {
+      "epoch": 0.38023952095808383,
+      "grad_norm": 1.0018425815907332,
+      "learning_rate": 9.134345191271742e-06,
+      "loss": 0.397,
+      "step": 127
+    },
+    {
+      "epoch": 0.38323353293413176,
+      "grad_norm": 0.8884698339357936,
+      "learning_rate": 9.121074858959997e-06,
+      "loss": 0.3779,
+      "step": 128
+    },
+    {
+      "epoch": 0.38622754491017963,
+      "grad_norm": 0.8022167109366266,
+      "learning_rate": 9.107713376679634e-06,
+      "loss": 0.3489,
+      "step": 129
+    },
+    {
+      "epoch": 0.38922155688622756,
+      "grad_norm": 0.9648398323735261,
+      "learning_rate": 9.094261039960028e-06,
+      "loss": 0.3578,
+      "step": 130
+    },
+    {
+      "epoch": 0.39221556886227543,
+      "grad_norm": 1.021544334412822,
+      "learning_rate": 9.08071814634008e-06,
+      "loss": 0.4094,
+      "step": 131
+    },
+    {
+      "epoch": 0.39520958083832336,
+      "grad_norm": 0.9388222677297999,
+      "learning_rate": 9.067084995361623e-06,
+      "loss": 0.3568,
+      "step": 132
+    },
+    {
+      "epoch": 0.39820359281437123,
+      "grad_norm": 0.9213582496355888,
+      "learning_rate": 9.053361888562807e-06,
+      "loss": 0.3911,
+      "step": 133
+    },
+    {
+      "epoch": 0.40119760479041916,
+      "grad_norm": 0.9070931123497717,
+      "learning_rate": 9.039549129471423e-06,
+      "loss": 0.3355,
+      "step": 134
+    },
+    {
+      "epoch": 0.4041916167664671,
+      "grad_norm": 0.7663830916226176,
+      "learning_rate": 9.025647023598196e-06,
+      "loss": 0.3138,
+      "step": 135
+    },
+    {
+      "epoch": 0.40718562874251496,
+      "grad_norm": 0.8506065165395951,
+      "learning_rate": 9.011655878430018e-06,
+      "loss": 0.3186,
+      "step": 136
+    },
+    {
+      "epoch": 0.4101796407185629,
+      "grad_norm": 1.013195726903395,
+      "learning_rate": 8.99757600342316e-06,
+      "loss": 0.3842,
+      "step": 137
+    },
+    {
+      "epoch": 0.41317365269461076,
+      "grad_norm": 0.9269440305974377,
+      "learning_rate": 8.983407709996415e-06,
+      "loss": 0.3682,
+      "step": 138
+    },
+    {
+      "epoch": 0.4161676646706587,
+      "grad_norm": 0.9287493222992738,
+      "learning_rate": 8.969151311524215e-06,
+      "loss": 0.3393,
+      "step": 139
+    },
+    {
+      "epoch": 0.41916167664670656,
+      "grad_norm": 0.9911779016436961,
+      "learning_rate": 8.954807123329703e-06,
+      "loss": 0.3674,
+      "step": 140
+    },
+    {
+      "epoch": 0.4221556886227545,
+      "grad_norm": 0.7616393001115792,
+      "learning_rate": 8.940375462677758e-06,
+      "loss": 0.2976,
+      "step": 141
+    },
+    {
+      "epoch": 0.4251497005988024,
+      "grad_norm": 0.8535317280369027,
+      "learning_rate": 8.92585664876797e-06,
+      "loss": 0.3382,
+      "step": 142
+    },
+    {
+      "epoch": 0.4281437125748503,
+      "grad_norm": 0.8078148654373282,
+      "learning_rate": 8.911251002727588e-06,
+      "loss": 0.3328,
+      "step": 143
+    },
+    {
+      "epoch": 0.4311377245508982,
+      "grad_norm": 0.8016111522171929,
+      "learning_rate": 8.896558847604414e-06,
+      "loss": 0.317,
+      "step": 144
+    },
+    {
+      "epoch": 0.4341317365269461,
+      "grad_norm": 0.8485527935175589,
+      "learning_rate": 8.881780508359661e-06,
+      "loss": 0.4161,
+      "step": 145
+    },
+    {
+      "epoch": 0.437125748502994,
+      "grad_norm": 1.0868267443809434,
+      "learning_rate": 8.86691631186076e-06,
+      "loss": 0.3443,
+      "step": 146
+    },
+    {
+      "epoch": 0.44011976047904194,
+      "grad_norm": 0.7355685063239832,
+      "learning_rate": 8.851966586874138e-06,
+      "loss": 0.3052,
+      "step": 147
+    },
+    {
+      "epoch": 0.4431137724550898,
+      "grad_norm": 0.8723074459225714,
+      "learning_rate": 8.836931664057935e-06,
+      "loss": 0.3301,
+      "step": 148
+    },
+    {
+      "epoch": 0.44610778443113774,
+      "grad_norm": 0.9422935022179557,
+      "learning_rate": 8.821811875954705e-06,
+      "loss": 0.354,
+      "step": 149
+    },
+    {
+      "epoch": 0.4491017964071856,
+      "grad_norm": 0.8642038045560563,
+      "learning_rate": 8.806607556984045e-06,
+      "loss": 0.2997,
+      "step": 150
+    },
+    {
+      "epoch": 0.45209580838323354,
+      "grad_norm": 0.8407656972433648,
+      "learning_rate": 8.791319043435213e-06,
+      "loss": 0.3402,
+      "step": 151
+    },
+    {
+      "epoch": 0.4550898203592814,
+      "grad_norm": 0.8018946809646184,
+      "learning_rate": 8.775946673459682e-06,
+      "loss": 0.3465,
+      "step": 152
+    },
+    {
+      "epoch": 0.45808383233532934,
+      "grad_norm": 0.8780480684120829,
+      "learning_rate": 8.76049078706366e-06,
+      "loss": 0.3181,
+      "step": 153
+    },
+    {
+      "epoch": 0.46107784431137727,
+      "grad_norm": 0.8621829816704457,
+      "learning_rate": 8.744951726100572e-06,
+      "loss": 0.3073,
+      "step": 154
+    },
+    {
+      "epoch": 0.46407185628742514,
+      "grad_norm": 0.9745447580705036,
+      "learning_rate": 8.729329834263503e-06,
+      "loss": 0.3351,
+      "step": 155
+    },
+    {
+      "epoch": 0.46706586826347307,
+      "grad_norm": 0.8994048708150972,
+      "learning_rate": 8.713625457077585e-06,
+      "loss": 0.3602,
+      "step": 156
+    },
+    {
+      "epoch": 0.47005988023952094,
+      "grad_norm": 0.8571469075006177,
+      "learning_rate": 8.697838941892371e-06,
+      "loss": 0.3206,
+      "step": 157
+    },
+    {
+      "epoch": 0.47305389221556887,
+      "grad_norm": 0.9010847756525124,
+      "learning_rate": 8.681970637874131e-06,
+      "loss": 0.3716,
+      "step": 158
+    },
+    {
+      "epoch": 0.47604790419161674,
+      "grad_norm": 0.9208985296526275,
+      "learning_rate": 8.666020895998154e-06,
+      "loss": 0.3451,
+      "step": 159
+    },
+    {
+      "epoch": 0.47904191616766467,
+      "grad_norm": 0.9327590019141727,
+      "learning_rate": 8.64999006904096e-06,
+      "loss": 0.3555,
+      "step": 160
+    },
+    {
+      "epoch": 0.4820359281437126,
+      "grad_norm": 0.936434139540589,
+      "learning_rate": 8.63387851157252e-06,
+      "loss": 0.3603,
+      "step": 161
+    },
+    {
+      "epoch": 0.48502994011976047,
+      "grad_norm": 0.8929422740312684,
+      "learning_rate": 8.617686579948396e-06,
+      "loss": 0.3036,
+      "step": 162
+    },
+    {
+      "epoch": 0.4880239520958084,
+      "grad_norm": 0.946236181581822,
+      "learning_rate": 8.60141463230187e-06,
+      "loss": 0.316,
+      "step": 163
+    },
+    {
+      "epoch": 0.49101796407185627,
+      "grad_norm": 0.9407961833230595,
+      "learning_rate": 8.585063028536015e-06,
+      "loss": 0.3996,
+      "step": 164
+    },
+    {
+      "epoch": 0.4940119760479042,
+      "grad_norm": 0.8738064243187453,
+      "learning_rate": 8.568632130315747e-06,
+      "loss": 0.3424,
+      "step": 165
+    },
+    {
+      "epoch": 0.49700598802395207,
+      "grad_norm": 0.8777689262622684,
+      "learning_rate": 8.552122301059807e-06,
+      "loss": 0.334,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.865110328632691,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.3227,
+      "step": 167
+    },
+    {
+      "epoch": 0.5029940119760479,
+      "grad_norm": 0.8794443278528441,
+      "learning_rate": 8.518867311836808e-06,
+      "loss": 0.3664,
+      "step": 168
+    },
+    {
+      "epoch": 0.5059880239520959,
+      "grad_norm": 0.9340014915251268,
+      "learning_rate": 8.502122887403882e-06,
+      "loss": 0.3759,
+      "step": 169
+    },
+    {
+      "epoch": 0.5089820359281437,
+      "grad_norm": 0.9445323902311572,
+      "learning_rate": 8.485301002987285e-06,
+      "loss": 0.2994,
+      "step": 170
+    },
+    {
+      "epoch": 0.5119760479041916,
+      "grad_norm": 0.8936966398490986,
+      "learning_rate": 8.468402030653598e-06,
+      "loss": 0.3841,
+      "step": 171
+    },
+    {
+      "epoch": 0.5149700598802395,
+      "grad_norm": 0.8302390316603054,
+      "learning_rate": 8.451426344174433e-06,
+      "loss": 0.3266,
+      "step": 172
+    },
+    {
+      "epoch": 0.5179640718562875,
+      "grad_norm": 0.9604064259018522,
+      "learning_rate": 8.434374319018165e-06,
+      "loss": 0.3952,
+      "step": 173
+    },
+    {
+      "epoch": 0.5209580838323353,
+      "grad_norm": 0.8232906684450247,
+      "learning_rate": 8.417246332341638e-06,
+      "loss": 0.3225,
+      "step": 174
+    },
+    {
+      "epoch": 0.5239520958083832,
+      "grad_norm": 0.8116589790256981,
+      "learning_rate": 8.4000427629818e-06,
+      "loss": 0.3385,
+      "step": 175
+    },
+    {
+      "epoch": 0.5269461077844312,
+      "grad_norm": 0.886421773500398,
+      "learning_rate": 8.382763991447344e-06,
+      "loss": 0.366,
+      "step": 176
+    },
+    {
+      "epoch": 0.5299401197604791,
+      "grad_norm": 0.9279916258861699,
+      "learning_rate": 8.365410399910287e-06,
+      "loss": 0.3463,
+      "step": 177
+    },
+    {
+      "epoch": 0.5329341317365269,
+      "grad_norm": 1.0192430861819075,
+      "learning_rate": 8.347982372197515e-06,
+      "loss": 0.3398,
+      "step": 178
+    },
+    {
+      "epoch": 0.5359281437125748,
+      "grad_norm": 0.8664905568270995,
+      "learning_rate": 8.33048029378229e-06,
+      "loss": 0.352,
+      "step": 179
+    },
+    {
+      "epoch": 0.5389221556886228,
+      "grad_norm": 0.7921524482429922,
+      "learning_rate": 8.312904551775731e-06,
+      "loss": 0.3078,
+      "step": 180
+    },
+    {
+      "epoch": 0.5419161676646707,
+      "grad_norm": 0.83331553442875,
+      "learning_rate": 8.295255534918249e-06,
+      "loss": 0.3445,
+      "step": 181
+    },
+    {
+      "epoch": 0.5449101796407185,
+      "grad_norm": 1.0257469835323392,
+      "learning_rate": 8.277533633570948e-06,
+      "loss": 0.3982,
+      "step": 182
+    },
+    {
+      "epoch": 0.5479041916167665,
+      "grad_norm": 0.8470156833032758,
+      "learning_rate": 8.25973923970699e-06,
+      "loss": 0.3648,
+      "step": 183
+    },
+    {
+      "epoch": 0.5508982035928144,
+      "grad_norm": 0.7975668659718067,
+      "learning_rate": 8.241872746902934e-06,
+      "loss": 0.3994,
+      "step": 184
+    },
+    {
+      "epoch": 0.5538922155688623,
+      "grad_norm": 0.8638109946726131,
+      "learning_rate": 8.223934550330015e-06,
+      "loss": 0.3167,
+      "step": 185
+    },
+    {
+      "epoch": 0.5568862275449101,
+      "grad_norm": 0.9154610778281713,
+      "learning_rate": 8.20592504674542e-06,
+      "loss": 0.3722,
+      "step": 186
+    },
+    {
+      "epoch": 0.5598802395209581,
+      "grad_norm": 0.7882458173582081,
+      "learning_rate": 8.187844634483495e-06,
+      "loss": 0.3761,
+      "step": 187
+    },
+    {
+      "epoch": 0.562874251497006,
+      "grad_norm": 1.0180358902255366,
+      "learning_rate": 8.16969371344696e-06,
+      "loss": 0.4305,
+      "step": 188
+    },
+    {
+      "epoch": 0.5658682634730539,
+      "grad_norm": 0.7709323802887873,
+      "learning_rate": 8.151472685098037e-06,
+      "loss": 0.3438,
+      "step": 189
+    },
+    {
+      "epoch": 0.5688622754491018,
+      "grad_norm": 0.8344500598340894,
+      "learning_rate": 8.13318195244958e-06,
+      "loss": 0.3228,
+      "step": 190
+    },
+    {
+      "epoch": 0.5718562874251497,
+      "grad_norm": 0.8352211070158856,
+      "learning_rate": 8.114821920056177e-06,
+      "loss": 0.3426,
+      "step": 191
+    },
+    {
+      "epoch": 0.5748502994011976,
+      "grad_norm": 0.8098442304130882,
+      "learning_rate": 8.096392994005177e-06,
+      "loss": 0.3657,
+      "step": 192
+    },
+    {
+      "epoch": 0.5778443113772455,
+      "grad_norm": 0.7430042712253846,
+      "learning_rate": 8.077895581907719e-06,
+      "loss": 0.3395,
+      "step": 193
+    },
+    {
+      "epoch": 0.5808383233532934,
+      "grad_norm": 0.7394722412885261,
+      "learning_rate": 8.059330092889724e-06,
+      "loss": 0.2955,
+      "step": 194
+    },
+    {
+      "epoch": 0.5838323353293413,
+      "grad_norm": 0.8250904355210983,
+      "learning_rate": 8.040696937582833e-06,
+      "loss": 0.3353,
+      "step": 195
+    },
+    {
+      "epoch": 0.5868263473053892,
+      "grad_norm": 0.9012937573479463,
+      "learning_rate": 8.021996528115335e-06,
+      "loss": 0.3136,
+      "step": 196
+    },
+    {
+      "epoch": 0.5898203592814372,
+      "grad_norm": 0.8755756618422285,
+      "learning_rate": 8.003229278103044e-06,
+      "loss": 0.3239,
+      "step": 197
+    },
+    {
+      "epoch": 0.592814371257485,
+      "grad_norm": 0.8659340320635494,
+      "learning_rate": 7.984395602640153e-06,
+      "loss": 0.3241,
+      "step": 198
+    },
+    {
+      "epoch": 0.5958083832335329,
+      "grad_norm": 0.9477354999471376,
+      "learning_rate": 7.96549591829006e-06,
+      "loss": 0.3421,
+      "step": 199
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "grad_norm": 0.7900942009412749,
+      "learning_rate": 7.946530643076138e-06,
+      "loss": 0.2907,
+      "step": 200
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "eval_loss": 0.33467286825180054,
+      "eval_runtime": 3.5746,
+      "eval_samples_per_second": 15.106,
+      "eval_steps_per_second": 3.916,
+      "step": 200
+    },
+    {
+      "epoch": 0.6017964071856288,
+      "grad_norm": 0.79312094170382,
+      "learning_rate": 7.927500196472506e-06,
+      "loss": 0.3338,
+      "step": 201
+    },
+    {
+      "epoch": 0.6047904191616766,
+      "grad_norm": 0.9034217570817616,
+      "learning_rate": 7.908404999394747e-06,
+      "loss": 0.2964,
+      "step": 202
+    },
+    {
+      "epoch": 0.6077844311377245,
+      "grad_norm": 0.8586808639213538,
+      "learning_rate": 7.889245474190588e-06,
+      "loss": 0.332,
+      "step": 203
+    },
+    {
+      "epoch": 0.6107784431137725,
+      "grad_norm": 1.0051484162950264,
+      "learning_rate": 7.870022044630569e-06,
+      "loss": 0.3651,
+      "step": 204
+    },
+    {
+      "epoch": 0.6137724550898204,
+      "grad_norm": 0.8213608666013195,
+      "learning_rate": 7.85073513589867e-06,
+      "loss": 0.3368,
+      "step": 205
+    },
+    {
+      "epoch": 0.6167664670658682,
+      "grad_norm": 1.0557670504490637,
+      "learning_rate": 7.831385174582901e-06,
+      "loss": 0.3467,
+      "step": 206
+    },
+    {
+      "epoch": 0.6197604790419161,
+      "grad_norm": 0.8305844226434226,
+      "learning_rate": 7.81197258866587e-06,
+      "loss": 0.3415,
+      "step": 207
+    },
+    {
+      "epoch": 0.6227544910179641,
+      "grad_norm": 0.8810023832436372,
+      "learning_rate": 7.792497807515317e-06,
+      "loss": 0.304,
+      "step": 208
+    },
+    {
+      "epoch": 0.625748502994012,
+      "grad_norm": 0.8855634435213009,
+      "learning_rate": 7.772961261874615e-06,
+      "loss": 0.2946,
+      "step": 209
+    },
+    {
+      "epoch": 0.6287425149700598,
+      "grad_norm": 0.9067453911759643,
+      "learning_rate": 7.75336338385325e-06,
+      "loss": 0.3253,
+      "step": 210
+    },
+    {
+      "epoch": 0.6317365269461078,
+      "grad_norm": 0.8898244780755367,
+      "learning_rate": 7.733704606917248e-06,
+      "loss": 0.3478,
+      "step": 211
+    },
+    {
+      "epoch": 0.6347305389221557,
+      "grad_norm": 0.8738916048358085,
+      "learning_rate": 7.713985365879607e-06,
+      "loss": 0.3054,
+      "step": 212
+    },
+    {
+      "epoch": 0.6377245508982036,
+      "grad_norm": 0.8500188297265527,
+      "learning_rate": 7.694206096890667e-06,
+      "loss": 0.3169,
+      "step": 213
+    },
+    {
+      "epoch": 0.6407185628742516,
+      "grad_norm": 0.844256505268219,
+      "learning_rate": 7.674367237428467e-06,
+      "loss": 0.3071,
+      "step": 214
+    },
+    {
+      "epoch": 0.6437125748502994,
+      "grad_norm": 0.8379775205343643,
+      "learning_rate": 7.654469226289068e-06,
+      "loss": 0.2913,
+      "step": 215
+    },
+    {
+      "epoch": 0.6467065868263473,
+      "grad_norm": 0.8767560769911583,
+      "learning_rate": 7.63451250357685e-06,
+      "loss": 0.3501,
+      "step": 216
+    },
+    {
+      "epoch": 0.6497005988023952,
+      "grad_norm": 0.896172615000714,
+      "learning_rate": 7.614497510694774e-06,
+      "loss": 0.3558,
+      "step": 217
+    },
+    {
+      "epoch": 0.6526946107784432,
+      "grad_norm": 0.9346507675555815,
+      "learning_rate": 7.5944246903346204e-06,
+      "loss": 0.3217,
+      "step": 218
+    },
+    {
+      "epoch": 0.655688622754491,
+      "grad_norm": 0.8375284268531287,
+      "learning_rate": 7.574294486467204e-06,
+      "loss": 0.3219,
+      "step": 219
+    },
+    {
+      "epoch": 0.6586826347305389,
+      "grad_norm": 0.8616776383120752,
+      "learning_rate": 7.55410734433254e-06,
+      "loss": 0.3532,
+      "step": 220
+    },
+    {
+      "epoch": 0.6616766467065869,
+      "grad_norm": 0.9431676744062615,
+      "learning_rate": 7.533863710430011e-06,
+      "loss": 0.3687,
+      "step": 221
+    },
+    {
+      "epoch": 0.6646706586826348,
+      "grad_norm": 0.8488729393047406,
+      "learning_rate": 7.513564032508484e-06,
+      "loss": 0.309,
+      "step": 222
+    },
+    {
+      "epoch": 0.6676646706586826,
+      "grad_norm": 0.9355238777931589,
+      "learning_rate": 7.493208759556406e-06,
+      "loss": 0.3665,
+      "step": 223
+    },
+    {
+      "epoch": 0.6706586826347305,
+      "grad_norm": 0.9566226952897219,
+      "learning_rate": 7.472798341791877e-06,
+      "loss": 0.3868,
+      "step": 224
+    },
+    {
+      "epoch": 0.6736526946107785,
+      "grad_norm": 0.8072744247437709,
+      "learning_rate": 7.452333230652688e-06,
+      "loss": 0.3211,
+      "step": 225
+    },
+    {
+      "epoch": 0.6766467065868264,
+      "grad_norm": 0.7841398004064947,
+      "learning_rate": 7.431813878786343e-06,
+      "loss": 0.3181,
+      "step": 226
+    },
+    {
+      "epoch": 0.6796407185628742,
+      "grad_norm": 0.8723747151692307,
+      "learning_rate": 7.4112407400400395e-06,
+      "loss": 0.3509,
+      "step": 227
+    },
+    {
+      "epoch": 0.6826347305389222,
+      "grad_norm": 0.8260131752149777,
+      "learning_rate": 7.390614269450633e-06,
+      "loss": 0.3255,
+      "step": 228
+    },
+    {
+      "epoch": 0.6856287425149701,
+      "grad_norm": 0.7681373368953692,
+      "learning_rate": 7.369934923234577e-06,
+      "loss": 0.3091,
+      "step": 229
+    },
+    {
+      "epoch": 0.688622754491018,
+      "grad_norm": 0.9535284481872933,
+      "learning_rate": 7.349203158777826e-06,
+      "loss": 0.3794,
+      "step": 230
+    },
+    {
+      "epoch": 0.6916167664670658,
+      "grad_norm": 0.916802050557078,
+      "learning_rate": 7.32841943462572e-06,
+      "loss": 0.3423,
+      "step": 231
+    },
+    {
+      "epoch": 0.6946107784431138,
+      "grad_norm": 0.9287242102831487,
+      "learning_rate": 7.3075842104728445e-06,
+      "loss": 0.3495,
+      "step": 232
+    },
+    {
+      "epoch": 0.6976047904191617,
+      "grad_norm": 0.8792795000931635,
+      "learning_rate": 7.286697947152868e-06,
+      "loss": 0.3601,
+      "step": 233
+    },
+    {
+      "epoch": 0.7005988023952096,
+      "grad_norm": 0.8669066098912112,
+      "learning_rate": 7.265761106628338e-06,
+      "loss": 0.3235,
+      "step": 234
+    },
+    {
+      "epoch": 0.7035928143712575,
+      "grad_norm": 0.8135169115612799,
+      "learning_rate": 7.244774151980466e-06,
+      "loss": 0.3155,
+      "step": 235
+    },
+    {
+      "epoch": 0.7065868263473054,
+      "grad_norm": 1.0900128075304087,
+      "learning_rate": 7.223737547398898e-06,
+      "loss": 0.3287,
+      "step": 236
+    },
+    {
+      "epoch": 0.7095808383233533,
+      "grad_norm": 0.8782965885774511,
+      "learning_rate": 7.20265175817143e-06,
+      "loss": 0.3342,
+      "step": 237
+    },
+    {
+      "epoch": 0.7125748502994012,
+      "grad_norm": 0.9486594176816813,
+      "learning_rate": 7.181517250673729e-06,
+      "loss": 0.3711,
+      "step": 238
+    },
+    {
+      "epoch": 0.7155688622754491,
+      "grad_norm": 0.9032070160794724,
+      "learning_rate": 7.1603344923590065e-06,
+      "loss": 0.3279,
+      "step": 239
+    },
+    {
+      "epoch": 0.718562874251497,
+      "grad_norm": 0.8552847592774973,
+      "learning_rate": 7.139103951747694e-06,
+      "loss": 0.3598,
+      "step": 240
+    },
+    {
+      "epoch": 0.7215568862275449,
+      "grad_norm": 0.8463195466291924,
+      "learning_rate": 7.1178260984170675e-06,
+      "loss": 0.3046,
+      "step": 241
+    },
+    {
+      "epoch": 0.7245508982035929,
+      "grad_norm": 0.8371218456503741,
+      "learning_rate": 7.0965014029908654e-06,
+      "loss": 0.3036,
+      "step": 242
+    },
+    {
+      "epoch": 0.7275449101796407,
+      "grad_norm": 0.7820672325449471,
+      "learning_rate": 7.075130337128883e-06,
+      "loss": 0.3335,
+      "step": 243
+    },
+    {
+      "epoch": 0.7305389221556886,
+      "grad_norm": 0.9192843234012044,
+      "learning_rate": 7.053713373516538e-06,
+      "loss": 0.3542,
+      "step": 244
+    },
+    {
+      "epoch": 0.7335329341317365,
+      "grad_norm": 0.8379640798614526,
+      "learning_rate": 7.03225098585441e-06,
+      "loss": 0.3584,
+      "step": 245
+    },
+    {
+      "epoch": 0.7365269461077845,
+      "grad_norm": 0.7781508272722515,
+      "learning_rate": 7.0107436488477694e-06,
+      "loss": 0.3443,
+      "step": 246
+    },
+    {
+      "epoch": 0.7395209580838323,
+      "grad_norm": 0.8834077256780015,
+      "learning_rate": 6.989191838196083e-06,
+      "loss": 0.3627,
+      "step": 247
+    },
+    {
+      "epoch": 0.7425149700598802,
+      "grad_norm": 0.8544321583408758,
+      "learning_rate": 6.9675960305824785e-06,
+      "loss": 0.3155,
+      "step": 248
+    },
+    {
+      "epoch": 0.7455089820359282,
+      "grad_norm": 0.9527967928819479,
+      "learning_rate": 6.945956703663212e-06,
+      "loss": 0.3541,
+      "step": 249
+    },
+    {
+      "epoch": 0.7485029940119761,
+      "grad_norm": 0.8271194232057484,
+      "learning_rate": 6.9242743360570985e-06,
+      "loss": 0.3229,
+      "step": 250
+    },
+    {
+      "epoch": 0.7514970059880239,
+      "grad_norm": 0.9601665766412376,
+      "learning_rate": 6.9025494073349284e-06,
+      "loss": 0.3676,
+      "step": 251
+    },
+    {
+      "epoch": 0.7544910179640718,
+      "grad_norm": 0.7929259708210948,
+      "learning_rate": 6.880782398008862e-06,
+      "loss": 0.3347,
+      "step": 252
+    },
+    {
+      "epoch": 0.7574850299401198,
+      "grad_norm": 0.887569273022383,
+      "learning_rate": 6.858973789521792e-06,
+      "loss": 0.3027,
+      "step": 253
+    },
+    {
+      "epoch": 0.7604790419161677,
+      "grad_norm": 0.9319346859656158,
+      "learning_rate": 6.837124064236709e-06,
+      "loss": 0.3641,
+      "step": 254
+    },
+    {
+      "epoch": 0.7634730538922155,
+      "grad_norm": 0.7943563622003768,
+      "learning_rate": 6.815233705426019e-06,
+      "loss": 0.3319,
+      "step": 255
+    },
+    {
+      "epoch": 0.7664670658682635,
+      "grad_norm": 0.9084023668834192,
+      "learning_rate": 6.7933031972608644e-06,
+      "loss": 0.3609,
+      "step": 256
+    },
+    {
+      "epoch": 0.7694610778443114,
+      "grad_norm": 0.8153357455537337,
+      "learning_rate": 6.771333024800411e-06,
+      "loss": 0.3456,
+      "step": 257
+    },
+    {
+      "epoch": 0.7724550898203593,
+      "grad_norm": 1.032838951696525,
+      "learning_rate": 6.74932367398112e-06,
+      "loss": 0.3308,
+      "step": 258
+    },
+    {
+      "epoch": 0.7754491017964071,
+      "grad_norm": 0.8492434450017045,
+      "learning_rate": 6.727275631605996e-06,
+      "loss": 0.2931,
+      "step": 259
+    },
+    {
+      "epoch": 0.7784431137724551,
+      "grad_norm": 0.9059912918790543,
+      "learning_rate": 6.70518938533383e-06,
+      "loss": 0.3325,
+      "step": 260
+    },
+    {
+      "epoch": 0.781437125748503,
+      "grad_norm": 0.9460828916774942,
+      "learning_rate": 6.683065423668403e-06,
+      "loss": 0.3502,
+      "step": 261
+    },
+    {
+      "epoch": 0.7844311377245509,
+      "grad_norm": 0.9364179318985469,
+      "learning_rate": 6.660904235947687e-06,
+      "loss": 0.3335,
+      "step": 262
+    },
+    {
+      "epoch": 0.7874251497005988,
+      "grad_norm": 0.8603373377298055,
+      "learning_rate": 6.638706312333018e-06,
+      "loss": 0.3519,
+      "step": 263
+    },
+    {
+      "epoch": 0.7904191616766467,
+      "grad_norm": 0.8914430701461508,
+      "learning_rate": 6.61647214379826e-06,
+      "loss": 0.3857,
+      "step": 264
+    },
+    {
+      "epoch": 0.7934131736526946,
+      "grad_norm": 0.9617998429624773,
+      "learning_rate": 6.594202222118941e-06,
+      "loss": 0.3498,
+      "step": 265
+    },
+    {
+      "epoch": 0.7964071856287425,
+      "grad_norm": 0.9106443714228667,
+      "learning_rate": 6.571897039861377e-06,
+      "loss": 0.3341,
+      "step": 266
+    },
+    {
+      "epoch": 0.7994011976047904,
+      "grad_norm": 0.8730506251318221,
+      "learning_rate": 6.549557090371775e-06,
+      "loss": 0.345,
+      "step": 267
+    },
+    {
+      "epoch": 0.8023952095808383,
+      "grad_norm": 0.801950451184709,
+      "learning_rate": 6.527182867765333e-06,
+      "loss": 0.2552,
+      "step": 268
+    },
+    {
+      "epoch": 0.8053892215568862,
+      "grad_norm": 0.935679283867518,
+      "learning_rate": 6.504774866915291e-06,
+      "loss": 0.3341,
+      "step": 269
+    },
+    {
+      "epoch": 0.8083832335329342,
+      "grad_norm": 0.9778825623559328,
+      "learning_rate": 6.482333583442002e-06,
+      "loss": 0.3018,
+      "step": 270
+    },
+    {
+      "epoch": 0.811377245508982,
+      "grad_norm": 0.7869918395866576,
+      "learning_rate": 6.459859513701967e-06,
+      "loss": 0.3345,
+      "step": 271
+    },
+    {
+      "epoch": 0.8143712574850299,
+      "grad_norm": 1.0031750015830145,
+      "learning_rate": 6.437353154776848e-06,
+      "loss": 0.3393,
+      "step": 272
+    },
+    {
+      "epoch": 0.8173652694610778,
+      "grad_norm": 0.9085710799739674,
+      "learning_rate": 6.414815004462483e-06,
+      "loss": 0.3687,
+      "step": 273
+    },
+    {
+      "epoch": 0.8203592814371258,
+      "grad_norm": 0.8621398941933154,
+      "learning_rate": 6.3922455612578715e-06,
+      "loss": 0.3623,
+      "step": 274
+    },
+    {
+      "epoch": 0.8233532934131736,
+      "grad_norm": 0.855228401264576,
+      "learning_rate": 6.369645324354149e-06,
+      "loss": 0.2856,
+      "step": 275
+    },
+    {
+      "epoch": 0.8263473053892215,
+      "grad_norm": 0.8076193648394628,
+      "learning_rate": 6.3470147936235485e-06,
+      "loss": 0.3406,
+      "step": 276
+    },
+    {
+      "epoch": 0.8293413173652695,
+      "grad_norm": 0.7245911322472699,
+      "learning_rate": 6.3243544696083355e-06,
+      "loss": 0.3002,
+      "step": 277
+    },
+    {
+      "epoch": 0.8323353293413174,
+      "grad_norm": 0.8280158469680207,
+      "learning_rate": 6.301664853509755e-06,
+      "loss": 0.2891,
+      "step": 278
+    },
+    {
+      "epoch": 0.8353293413173652,
+      "grad_norm": 0.7854270076254557,
+      "learning_rate": 6.278946447176924e-06,
+      "loss": 0.3257,
+      "step": 279
+    },
+    {
+      "epoch": 0.8383233532934131,
+      "grad_norm": 0.7761777582444528,
+      "learning_rate": 6.256199753095745e-06,
+      "loss": 0.3063,
+      "step": 280
+    },
+    {
+      "epoch": 0.8413173652694611,
+      "grad_norm": 0.842302593081833,
+      "learning_rate": 6.233425274377793e-06,
+      "loss": 0.3318,
+      "step": 281
+    },
+    {
+      "epoch": 0.844311377245509,
+      "grad_norm": 1.0077950393614565,
+      "learning_rate": 6.21062351474918e-06,
+      "loss": 0.3674,
+      "step": 282
+    },
+    {
+      "epoch": 0.8473053892215568,
+      "grad_norm": 0.8701693556750275,
+      "learning_rate": 6.18779497853942e-06,
+      "loss": 0.3545,
+      "step": 283
+    },
+    {
+      "epoch": 0.8502994011976048,
+      "grad_norm": 0.8772047033068309,
+      "learning_rate": 6.164940170670266e-06,
+      "loss": 0.3602,
+      "step": 284
+    },
+    {
+      "epoch": 0.8532934131736527,
+      "grad_norm": 0.7301435777944242,
+      "learning_rate": 6.142059596644557e-06,
+      "loss": 0.2833,
+      "step": 285
+    },
+    {
+      "epoch": 0.8562874251497006,
+      "grad_norm": 0.8481337976116274,
+      "learning_rate": 6.11915376253502e-06,
+      "loss": 0.3102,
+      "step": 286
+    },
+    {
+      "epoch": 0.8592814371257484,
+      "grad_norm": 0.8863490665161842,
+      "learning_rate": 6.096223174973091e-06,
+      "loss": 0.3412,
+      "step": 287
+    },
+    {
+      "epoch": 0.8622754491017964,
+      "grad_norm": 0.7692810095706181,
+      "learning_rate": 6.073268341137694e-06,
+      "loss": 0.2856,
+      "step": 288
+    },
+    {
+      "epoch": 0.8652694610778443,
+      "grad_norm": 0.8876856270528781,
+      "learning_rate": 6.050289768744042e-06,
+      "loss": 0.2781,
+      "step": 289
+    },
+    {
+      "epoch": 0.8682634730538922,
+      "grad_norm": 0.9655511225084505,
+      "learning_rate": 6.0272879660323936e-06,
+      "loss": 0.3611,
+      "step": 290
+    },
+    {
+      "epoch": 0.8712574850299402,
+      "grad_norm": 0.8801553905843293,
+      "learning_rate": 6.004263441756815e-06,
+      "loss": 0.3386,
+      "step": 291
+    },
+    {
+      "epoch": 0.874251497005988,
+      "grad_norm": 0.8644727105745302,
+      "learning_rate": 5.98121670517393e-06,
+      "loss": 0.3101,
+      "step": 292
+    },
+    {
+      "epoch": 0.8772455089820359,
+      "grad_norm": 0.7541522693012973,
+      "learning_rate": 5.958148266031654e-06,
+      "loss": 0.2861,
+      "step": 293
+    },
+    {
+      "epoch": 0.8802395209580839,
+      "grad_norm": 0.8045210692490931,
+      "learning_rate": 5.935058634557917e-06,
+      "loss": 0.3023,
+      "step": 294
+    },
+    {
+      "epoch": 0.8832335329341318,
+      "grad_norm": 0.7439806112500583,
+      "learning_rate": 5.911948321449384e-06,
+      "loss": 0.3294,
+      "step": 295
+    },
+    {
+      "epoch": 0.8862275449101796,
+      "grad_norm": 0.8858437389929406,
+      "learning_rate": 5.8888178378601565e-06,
+      "loss": 0.3423,
+      "step": 296
+    },
+    {
+      "epoch": 0.8892215568862275,
+      "grad_norm": 0.9221571927934884,
+      "learning_rate": 5.865667695390468e-06,
+      "loss": 0.3243,
+      "step": 297
+    },
+    {
+      "epoch": 0.8922155688622755,
+      "grad_norm": 0.8235617800073382,
+      "learning_rate": 5.842498406075363e-06,
+      "loss": 0.3161,
+      "step": 298
+    },
+    {
+      "epoch": 0.8952095808383234,
+      "grad_norm": 0.8975683741325864,
+      "learning_rate": 5.819310482373381e-06,
+      "loss": 0.3295,
+      "step": 299
+    },
+    {
+      "epoch": 0.8982035928143712,
+      "grad_norm": 0.849503595809189,
+      "learning_rate": 5.796104437155213e-06,
+      "loss": 0.3629,
+      "step": 300
+    },
+    {
+      "epoch": 0.9011976047904192,
+      "grad_norm": 0.8681035836027373,
+      "learning_rate": 5.772880783692363e-06,
+      "loss": 0.309,
+      "step": 301
+    },
+    {
+      "epoch": 0.9041916167664671,
+      "grad_norm": 0.940434193489446,
+      "learning_rate": 5.749640035645798e-06,
+      "loss": 0.3,
+      "step": 302
+    },
+    {
+      "epoch": 0.907185628742515,
+      "grad_norm": 0.9053734789495154,
+      "learning_rate": 5.726382707054578e-06,
+      "loss": 0.3286,
+      "step": 303
+    },
+    {
+      "epoch": 0.9101796407185628,
+      "grad_norm": 0.9221578596023046,
+      "learning_rate": 5.703109312324493e-06,
+      "loss": 0.3232,
+      "step": 304
+    },
+    {
+      "epoch": 0.9131736526946108,
+      "grad_norm": 0.9396431013868058,
+      "learning_rate": 5.679820366216684e-06,
+      "loss": 0.3874,
+      "step": 305
+    },
+    {
+      "epoch": 0.9161676646706587,
+      "grad_norm": 0.7994743476501736,
+      "learning_rate": 5.656516383836263e-06,
+      "loss": 0.2814,
+      "step": 306
+    },
+    {
+      "epoch": 0.9191616766467066,
+      "grad_norm": 0.8469001159251874,
+      "learning_rate": 5.6331978806209044e-06,
+      "loss": 0.3352,
+      "step": 307
+    },
+    {
+      "epoch": 0.9221556886227545,
+      "grad_norm": 0.884067560153009,
+      "learning_rate": 5.609865372329461e-06,
+      "loss": 0.3084,
+      "step": 308
+    },
+    {
+      "epoch": 0.9251497005988024,
+      "grad_norm": 0.8169845922748498,
+      "learning_rate": 5.586519375030549e-06,
+      "loss": 0.2773,
+      "step": 309
+    },
+    {
+      "epoch": 0.9281437125748503,
+      "grad_norm": 0.7863783585710435,
+      "learning_rate": 5.5631604050911354e-06,
+      "loss": 0.3033,
+      "step": 310
+    },
+    {
+      "epoch": 0.9311377245508982,
+      "grad_norm": 0.948623009920141,
+      "learning_rate": 5.539788979165115e-06,
+      "loss": 0.2917,
+      "step": 311
+    },
+    {
+      "epoch": 0.9341317365269461,
+      "grad_norm": 0.779757317416599,
+      "learning_rate": 5.516405614181883e-06,
+      "loss": 0.2739,
+      "step": 312
+    },
+    {
+      "epoch": 0.937125748502994,
+      "grad_norm": 0.7695694632358886,
+      "learning_rate": 5.4930108273349034e-06,
+      "loss": 0.306,
+      "step": 313
+    },
+    {
+      "epoch": 0.9401197604790419,
+      "grad_norm": 0.8522371811136099,
+      "learning_rate": 5.4696051360702725e-06,
+      "loss": 0.2925,
+      "step": 314
+    },
+    {
+      "epoch": 0.9431137724550899,
+      "grad_norm": 0.9537835504530565,
+      "learning_rate": 5.446189058075265e-06,
+      "loss": 0.3472,
+      "step": 315
+    },
+    {
+      "epoch": 0.9461077844311377,
+      "grad_norm": 0.8724112927980736,
+      "learning_rate": 5.4227631112668955e-06,
+      "loss": 0.3496,
+      "step": 316
+    },
+    {
+      "epoch": 0.9491017964071856,
+      "grad_norm": 0.9143153514381644,
+      "learning_rate": 5.39932781378045e-06,
+      "loss": 0.3519,
+      "step": 317
+    },
+    {
+      "epoch": 0.9520958083832335,
+      "grad_norm": 0.7615328536144775,
+      "learning_rate": 5.375883683958041e-06,
+      "loss": 0.3098,
+      "step": 318
+    },
+    {
+      "epoch": 0.9550898203592815,
+      "grad_norm": 0.8649703731317651,
+      "learning_rate": 5.3524312403371255e-06,
+      "loss": 0.325,
+      "step": 319
+    },
+    {
+      "epoch": 0.9580838323353293,
+      "grad_norm": 0.9723484494302708,
+      "learning_rate": 5.328971001639054e-06,
+      "loss": 0.291,
+      "step": 320
+    },
+    {
+      "epoch": 0.9610778443113772,
+      "grad_norm": 0.8490247110632723,
+      "learning_rate": 5.3055034867575825e-06,
+      "loss": 0.2935,
+      "step": 321
+    },
+    {
+      "epoch": 0.9640718562874252,
+      "grad_norm": 0.9988398704436575,
+      "learning_rate": 5.282029214747404e-06,
+      "loss": 0.3491,
+      "step": 322
+    },
+    {
+      "epoch": 0.9670658682634731,
+      "grad_norm": 0.8543590517207817,
+      "learning_rate": 5.258548704812667e-06,
+      "loss": 0.3286,
+      "step": 323
+    },
+    {
+      "epoch": 0.9700598802395209,
+      "grad_norm": 0.8500233833887514,
+      "learning_rate": 5.235062476295488e-06,
+      "loss": 0.3151,
+      "step": 324
+    },
+    {
+      "epoch": 0.9730538922155688,
+      "grad_norm": 0.8765244333310814,
+      "learning_rate": 5.211571048664469e-06,
+      "loss": 0.2808,
+      "step": 325
+    },
+    {
+      "epoch": 0.9760479041916168,
+      "grad_norm": 0.9333931872302441,
+      "learning_rate": 5.188074941503203e-06,
+      "loss": 0.3391,
+      "step": 326
+    },
+    {
+      "epoch": 0.9790419161676647,
+      "grad_norm": 0.8773054912858921,
+      "learning_rate": 5.164574674498788e-06,
+      "loss": 0.3265,
+      "step": 327
+    },
+    {
+      "epoch": 0.9820359281437125,
+      "grad_norm": 0.9665585824211138,
+      "learning_rate": 5.141070767430331e-06,
+      "loss": 0.3117,
+      "step": 328
+    },
+    {
+      "epoch": 0.9850299401197605,
+      "grad_norm": 0.8395082187919094,
+      "learning_rate": 5.117563740157444e-06,
+      "loss": 0.3105,
+      "step": 329
+    },
+    {
+      "epoch": 0.9880239520958084,
+      "grad_norm": 0.9568305242413174,
+      "learning_rate": 5.094054112608758e-06,
+      "loss": 0.3738,
+      "step": 330
+    },
+    {
+      "epoch": 0.9910179640718563,
+      "grad_norm": 0.7907755067956981,
+      "learning_rate": 5.070542404770413e-06,
+      "loss": 0.3541,
+      "step": 331
+    },
+    {
+      "epoch": 0.9940119760479041,
+      "grad_norm": 0.7841586952208952,
+      "learning_rate": 5.047029136674563e-06,
+      "loss": 0.2886,
+      "step": 332
+    },
+    {
+      "epoch": 0.9970059880239521,
+      "grad_norm": 0.9757677527538441,
+      "learning_rate": 5.023514828387868e-06,
+      "loss": 0.4342,
+      "step": 333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.843376725475,
+      "learning_rate": 5e-06,
+      "loss": 0.2437,
+      "step": 334
+    },
+    {
+      "epoch": 1.0029940119760479,
+      "grad_norm": 0.9097444729914339,
+      "learning_rate": 4.976485171612134e-06,
+      "loss": 0.204,
+      "step": 335
+    },
+    {
+      "epoch": 1.0059880239520957,
+      "grad_norm": 0.7545799997722877,
+      "learning_rate": 4.95297086332544e-06,
+      "loss": 0.2173,
+      "step": 336
+    },
+    {
+      "epoch": 1.0089820359281436,
+      "grad_norm": 0.8688981009379331,
+      "learning_rate": 4.9294575952295896e-06,
+      "loss": 0.2283,
+      "step": 337
+    },
+    {
+      "epoch": 1.0119760479041917,
+      "grad_norm": 0.8948621203651169,
+      "learning_rate": 4.905945887391242e-06,
+      "loss": 0.2547,
+      "step": 338
+    },
+    {
+      "epoch": 1.0149700598802396,
+      "grad_norm": 0.8942361934728625,
+      "learning_rate": 4.882436259842556e-06,
+      "loss": 0.2305,
+      "step": 339
+    },
+    {
+      "epoch": 1.0179640718562875,
+      "grad_norm": 0.8198085683609914,
+      "learning_rate": 4.858929232569671e-06,
+      "loss": 0.2231,
+      "step": 340
+    },
+    {
+      "epoch": 1.0209580838323353,
+      "grad_norm": 0.8621368121736944,
+      "learning_rate": 4.835425325501214e-06,
+      "loss": 0.2397,
+      "step": 341
+    },
+    {
+      "epoch": 1.0239520958083832,
+      "grad_norm": 0.7630809423349001,
+      "learning_rate": 4.811925058496799e-06,
+      "loss": 0.1838,
+      "step": 342
+    },
+    {
+      "epoch": 1.026946107784431,
+      "grad_norm": 0.9879337199392489,
+      "learning_rate": 4.788428951335534e-06,
+      "loss": 0.2453,
+      "step": 343
+    },
+    {
+      "epoch": 1.029940119760479,
+      "grad_norm": 0.7847842378029675,
+      "learning_rate": 4.7649375237045135e-06,
+      "loss": 0.2138,
+      "step": 344
+    },
+    {
+      "epoch": 1.032934131736527,
+      "grad_norm": 0.8485728075701262,
+      "learning_rate": 4.741451295187333e-06,
+      "loss": 0.1931,
+      "step": 345
+    },
+    {
+      "epoch": 1.035928143712575,
+      "grad_norm": 0.8454857994536729,
+      "learning_rate": 4.717970785252596e-06,
+      "loss": 0.2547,
+      "step": 346
+    },
+    {
+      "epoch": 1.0389221556886228,
+      "grad_norm": 0.8807215076735381,
+      "learning_rate": 4.694496513242418e-06,
+      "loss": 0.2497,
+      "step": 347
+    },
+    {
+      "epoch": 1.0419161676646707,
+      "grad_norm": 0.9629588475475359,
+      "learning_rate": 4.671028998360947e-06,
+      "loss": 0.2017,
+      "step": 348
+    },
+    {
+      "epoch": 1.0449101796407185,
+      "grad_norm": 0.9289149674759557,
+      "learning_rate": 4.647568759662876e-06,
+      "loss": 0.2513,
+      "step": 349
+    },
+    {
+      "epoch": 1.0479041916167664,
+      "grad_norm": 0.8762206020145612,
+      "learning_rate": 4.624116316041962e-06,
+      "loss": 0.2309,
+      "step": 350
+    },
+    {
+      "epoch": 1.0508982035928143,
+      "grad_norm": 0.8520308484249168,
+      "learning_rate": 4.600672186219551e-06,
+      "loss": 0.2362,
+      "step": 351
+    },
+    {
+      "epoch": 1.0538922155688624,
+      "grad_norm": 0.9933722610372858,
+      "learning_rate": 4.5772368887331044e-06,
+      "loss": 0.2676,
+      "step": 352
+    },
+    {
+      "epoch": 1.0568862275449102,
+      "grad_norm": 0.9040467210370616,
+      "learning_rate": 4.553810941924735e-06,
+      "loss": 0.2599,
+      "step": 353
+    },
+    {
+      "epoch": 1.0598802395209581,
+      "grad_norm": 0.9016868832745932,
+      "learning_rate": 4.530394863929728e-06,
+      "loss": 0.2274,
+      "step": 354
+    },
+    {
+      "epoch": 1.062874251497006,
+      "grad_norm": 0.900854863732707,
+      "learning_rate": 4.506989172665097e-06,
+      "loss": 0.2122,
+      "step": 355
+    },
+    {
+      "epoch": 1.0658682634730539,
+      "grad_norm": 0.9050002807028359,
+      "learning_rate": 4.483594385818119e-06,
+      "loss": 0.2121,
+      "step": 356
+    },
+    {
+      "epoch": 1.0688622754491017,
+      "grad_norm": 0.9886008436717765,
+      "learning_rate": 4.460211020834887e-06,
+      "loss": 0.234,
+      "step": 357
+    },
+    {
+      "epoch": 1.0718562874251496,
+      "grad_norm": 0.9669754306017594,
+      "learning_rate": 4.436839594908866e-06,
+      "loss": 0.2816,
+      "step": 358
+    },
+    {
+      "epoch": 1.0748502994011977,
+      "grad_norm": 0.9573444964703198,
+      "learning_rate": 4.4134806249694514e-06,
+      "loss": 0.2493,
+      "step": 359
+    },
+    {
+      "epoch": 1.0778443113772456,
+      "grad_norm": 0.9676254717641087,
+      "learning_rate": 4.39013462767054e-06,
+      "loss": 0.2713,
+      "step": 360
+    },
+    {
+      "epoch": 1.0808383233532934,
+      "grad_norm": 0.8349788466453513,
+      "learning_rate": 4.366802119379098e-06,
+      "loss": 0.1791,
+      "step": 361
+    },
+    {
+      "epoch": 1.0838323353293413,
+      "grad_norm": 0.8154007400233115,
+      "learning_rate": 4.34348361616374e-06,
+      "loss": 0.2653,
+      "step": 362
+    },
+    {
+      "epoch": 1.0868263473053892,
+      "grad_norm": 1.0043377603516084,
+      "learning_rate": 4.3201796337833165e-06,
+      "loss": 0.2542,
+      "step": 363
+    },
+    {
+      "epoch": 1.089820359281437,
+      "grad_norm": 1.0642550736938516,
+      "learning_rate": 4.29689068767551e-06,
+      "loss": 0.2753,
+      "step": 364
+    },
+    {
+      "epoch": 1.092814371257485,
+      "grad_norm": 0.8601084168362063,
+      "learning_rate": 4.273617292945425e-06,
+      "loss": 0.2268,
+      "step": 365
+    },
+    {
+      "epoch": 1.095808383233533,
+      "grad_norm": 0.8241752489029223,
+      "learning_rate": 4.250359964354203e-06,
+      "loss": 0.2391,
+      "step": 366
+    },
+    {
+      "epoch": 1.098802395209581,
+      "grad_norm": 0.963344944682232,
+      "learning_rate": 4.227119216307637e-06,
+      "loss": 0.2378,
+      "step": 367
+    },
+    {
+      "epoch": 1.1017964071856288,
+      "grad_norm": 0.8948262875622087,
+      "learning_rate": 4.203895562844789e-06,
+      "loss": 0.2707,
+      "step": 368
+    },
+    {
+      "epoch": 1.1047904191616766,
+      "grad_norm": 0.9738213077981859,
+      "learning_rate": 4.18068951762662e-06,
+      "loss": 0.2665,
+      "step": 369
+    },
+    {
+      "epoch": 1.1077844311377245,
+      "grad_norm": 0.8637469614510346,
+      "learning_rate": 4.157501593924638e-06,
+      "loss": 0.2149,
+      "step": 370
+    },
+    {
+      "epoch": 1.1107784431137724,
+      "grad_norm": 0.8459074254573029,
+      "learning_rate": 4.134332304609533e-06,
+      "loss": 0.2534,
+      "step": 371
+    },
+    {
+      "epoch": 1.1137724550898203,
+      "grad_norm": 0.7160677188052542,
+      "learning_rate": 4.111182162139844e-06,
+      "loss": 0.2256,
+      "step": 372
+    },
+    {
+      "epoch": 1.1167664670658684,
+      "grad_norm": 0.9583456205987185,
+      "learning_rate": 4.088051678550617e-06,
+      "loss": 0.2191,
+      "step": 373
+    },
+    {
+      "epoch": 1.1197604790419162,
+      "grad_norm": 0.9081389134137872,
+      "learning_rate": 4.064941365442084e-06,
+      "loss": 0.2371,
+      "step": 374
+    },
+    {
+      "epoch": 1.122754491017964,
+      "grad_norm": 1.0314125965352072,
+      "learning_rate": 4.041851733968348e-06,
+      "loss": 0.2726,
+      "step": 375
+    },
+    {
+      "epoch": 1.125748502994012,
+      "grad_norm": 0.9534911581960199,
+      "learning_rate": 4.018783294826071e-06,
+      "loss": 0.255,
+      "step": 376
+    },
+    {
+      "epoch": 1.1287425149700598,
+      "grad_norm": 0.9956615113691805,
+      "learning_rate": 3.995736558243186e-06,
+      "loss": 0.2696,
+      "step": 377
+    },
+    {
+      "epoch": 1.1317365269461077,
+      "grad_norm": 0.9006877062415116,
+      "learning_rate": 3.972712033967608e-06,
+      "loss": 0.2354,
+      "step": 378
+    },
+    {
+      "epoch": 1.1347305389221556,
+      "grad_norm": 0.9292170195044273,
+      "learning_rate": 3.949710231255961e-06,
+      "loss": 0.2796,
+      "step": 379
+    },
+    {
+      "epoch": 1.1377245508982037,
+      "grad_norm": 1.0526665863211797,
+      "learning_rate": 3.926731658862307e-06,
+      "loss": 0.2413,
+      "step": 380
+    },
+    {
+      "epoch": 1.1407185628742516,
+      "grad_norm": 0.9004024526743142,
+      "learning_rate": 3.903776825026912e-06,
+      "loss": 0.2643,
+      "step": 381
+    },
+    {
+      "epoch": 1.1437125748502994,
+      "grad_norm": 0.9130898278860868,
+      "learning_rate": 3.8808462374649805e-06,
+      "loss": 0.2489,
+      "step": 382
+    },
+    {
+      "epoch": 1.1467065868263473,
+      "grad_norm": 0.8928230542508337,
+      "learning_rate": 3.857940403355444e-06,
+      "loss": 0.2127,
+      "step": 383
+    },
+    {
+      "epoch": 1.1497005988023952,
+      "grad_norm": 0.9327282858724195,
+      "learning_rate": 3.8350598293297345e-06,
+      "loss": 0.2618,
+      "step": 384
+    },
+    {
+      "epoch": 1.152694610778443,
+      "grad_norm": 0.9281682264649125,
+      "learning_rate": 3.8122050214605822e-06,
+      "loss": 0.2226,
+      "step": 385
+    },
+    {
+      "epoch": 1.1556886227544911,
+      "grad_norm": 0.8822237439841015,
+      "learning_rate": 3.7893764852508207e-06,
+      "loss": 0.2491,
+      "step": 386
+    },
+    {
+      "epoch": 1.158682634730539,
+      "grad_norm": 0.8388865603712542,
+      "learning_rate": 3.766574725622208e-06,
+      "loss": 0.2168,
+      "step": 387
+    },
+    {
+      "epoch": 1.1616766467065869,
+      "grad_norm": 0.848960789892097,
+      "learning_rate": 3.7438002469042567e-06,
+      "loss": 0.2351,
+      "step": 388
+    },
+    {
+      "epoch": 1.1646706586826348,
+      "grad_norm": 0.9716696256710403,
+      "learning_rate": 3.721053552823078e-06,
+      "loss": 0.2413,
+      "step": 389
+    },
+    {
+      "epoch": 1.1676646706586826,
+      "grad_norm": 0.9487172677318534,
+      "learning_rate": 3.698335146490246e-06,
+      "loss": 0.2326,
+      "step": 390
+    },
+    {
+      "epoch": 1.1706586826347305,
+      "grad_norm": 1.0407935865632547,
+      "learning_rate": 3.675645530391665e-06,
+      "loss": 0.2244,
+      "step": 391
+    },
+    {
+      "epoch": 1.1736526946107784,
+      "grad_norm": 0.925397327810438,
+      "learning_rate": 3.652985206376455e-06,
+      "loss": 0.2125,
+      "step": 392
+    },
+    {
+      "epoch": 1.1766467065868262,
+      "grad_norm": 0.7389453949674737,
+      "learning_rate": 3.630354675645853e-06,
+      "loss": 0.2058,
+      "step": 393
+    },
+    {
+      "epoch": 1.1796407185628743,
+      "grad_norm": 0.9303186413078484,
+      "learning_rate": 3.6077544387421293e-06,
+      "loss": 0.2497,
+      "step": 394
+    },
+    {
+      "epoch": 1.1826347305389222,
+      "grad_norm": 0.9429611529622871,
+      "learning_rate": 3.5851849955375177e-06,
+      "loss": 0.2042,
+      "step": 395
+    },
+    {
+      "epoch": 1.18562874251497,
+      "grad_norm": 0.9401869819479897,
+      "learning_rate": 3.5626468452231534e-06,
+      "loss": 0.2209,
+      "step": 396
+    },
+    {
+      "epoch": 1.188622754491018,
+      "grad_norm": 0.8992896025621585,
+      "learning_rate": 3.540140486298035e-06,
+      "loss": 0.1849,
+      "step": 397
+    },
+    {
+      "epoch": 1.1916167664670658,
+      "grad_norm": 0.8423693426005414,
+      "learning_rate": 3.517666416557999e-06,
+      "loss": 0.2663,
+      "step": 398
+    },
+    {
+      "epoch": 1.1946107784431137,
+      "grad_norm": 0.9083761337895103,
+      "learning_rate": 3.495225133084712e-06,
+      "loss": 0.2327,
+      "step": 399
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "grad_norm": 0.8230200891995576,
+      "learning_rate": 3.472817132234669e-06,
+      "loss": 0.2535,
+      "step": 400
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "eval_loss": 0.33559077978134155,
+      "eval_runtime": 3.5782,
+      "eval_samples_per_second": 15.091,
+      "eval_steps_per_second": 3.913,
+      "step": 400
+    },
+    {
+      "epoch": 1.2005988023952097,
+      "grad_norm": 0.8934802983025646,
+      "learning_rate": 3.4504429096282246e-06,
+      "loss": 0.2663,
+      "step": 401
+    },
+    {
+      "epoch": 1.2035928143712575,
+      "grad_norm": 0.9089212009147788,
+      "learning_rate": 3.428102960138625e-06,
+      "loss": 0.2562,
+      "step": 402
+    },
+    {
+      "epoch": 1.2065868263473054,
+      "grad_norm": 0.9694772874115569,
+      "learning_rate": 3.405797777881059e-06,
+      "loss": 0.2645,
+      "step": 403
+    },
+    {
+      "epoch": 1.2095808383233533,
+      "grad_norm": 0.9510679768838551,
+      "learning_rate": 3.3835278562017405e-06,
+      "loss": 0.2195,
+      "step": 404
+    },
+    {
+      "epoch": 1.2125748502994012,
+      "grad_norm": 0.8638789626469725,
+      "learning_rate": 3.3612936876669834e-06,
+      "loss": 0.2236,
+      "step": 405
+    },
+    {
+      "epoch": 1.215568862275449,
+      "grad_norm": 0.8136320048040392,
+      "learning_rate": 3.3390957640523147e-06,
+      "loss": 0.2199,
+      "step": 406
+    },
+    {
+      "epoch": 1.218562874251497,
+      "grad_norm": 0.8399432142688728,
+      "learning_rate": 3.3169345763315986e-06,
+      "loss": 0.236,
+      "step": 407
+    },
+    {
+      "epoch": 1.221556886227545,
+      "grad_norm": 0.895132007721184,
+      "learning_rate": 3.29481061466617e-06,
+      "loss": 0.2506,
+      "step": 408
+    },
+    {
+      "epoch": 1.2245508982035929,
+      "grad_norm": 1.0315751441949677,
+      "learning_rate": 3.2727243683940045e-06,
+      "loss": 0.2336,
+      "step": 409
+    },
+    {
+      "epoch": 1.2275449101796407,
+      "grad_norm": 0.9294994902369398,
+      "learning_rate": 3.2506763260188824e-06,
+      "loss": 0.2569,
+      "step": 410
+    },
+    {
+      "epoch": 1.2305389221556886,
+      "grad_norm": 0.856843897399255,
+      "learning_rate": 3.2286669751995905e-06,
+      "loss": 0.2375,
+      "step": 411
+    },
+    {
+      "epoch": 1.2335329341317365,
+      "grad_norm": 0.981487886083542,
+      "learning_rate": 3.2066968027391377e-06,
+      "loss": 0.2418,
+      "step": 412
+    },
+    {
+      "epoch": 1.2365269461077844,
+      "grad_norm": 0.8363219451535354,
+      "learning_rate": 3.1847662945739833e-06,
+      "loss": 0.2281,
+      "step": 413
+    },
+    {
+      "epoch": 1.2395209580838324,
+      "grad_norm": 0.9293679889082452,
+      "learning_rate": 3.1628759357632943e-06,
+      "loss": 0.2467,
+      "step": 414
+    },
+    {
+      "epoch": 1.2425149700598803,
+      "grad_norm": 1.0280751031518225,
+      "learning_rate": 3.1410262104782086e-06,
+      "loss": 0.2172,
+      "step": 415
+    },
+    {
+      "epoch": 1.2455089820359282,
+      "grad_norm": 0.9800325945674049,
+      "learning_rate": 3.119217601991139e-06,
+      "loss": 0.2333,
+      "step": 416
+    },
+    {
+      "epoch": 1.248502994011976,
+      "grad_norm": 0.9889659524131217,
+      "learning_rate": 3.0974505926650724e-06,
+      "loss": 0.209,
+      "step": 417
+    },
+    {
+      "epoch": 1.251497005988024,
+      "grad_norm": 0.8401195322703149,
+      "learning_rate": 3.0757256639429027e-06,
+      "loss": 0.2328,
+      "step": 418
+    },
+    {
+      "epoch": 1.2544910179640718,
+      "grad_norm": 0.9467402067032646,
+      "learning_rate": 3.0540432963367907e-06,
+      "loss": 0.2626,
+      "step": 419
+    },
+    {
+      "epoch": 1.2574850299401197,
+      "grad_norm": 0.9149526790168448,
+      "learning_rate": 3.032403969417523e-06,
+      "loss": 0.2266,
+      "step": 420
+    },
+    {
+      "epoch": 1.2604790419161676,
+      "grad_norm": 0.8399102905235053,
+      "learning_rate": 3.010808161803917e-06,
+      "loss": 0.2182,
+      "step": 421
+    },
+    {
+      "epoch": 1.2634730538922156,
+      "grad_norm": 0.8572625991470963,
+      "learning_rate": 2.9892563511522305e-06,
+      "loss": 0.2541,
+      "step": 422
+    },
+    {
+      "epoch": 1.2664670658682635,
+      "grad_norm": 0.8878535847205529,
+      "learning_rate": 2.9677490141455915e-06,
+      "loss": 0.2121,
+      "step": 423
+    },
+    {
+      "epoch": 1.2694610778443114,
+      "grad_norm": 0.8984661284376178,
+      "learning_rate": 2.946286626483463e-06,
+      "loss": 0.2653,
+      "step": 424
+    },
+    {
+      "epoch": 1.2724550898203593,
+      "grad_norm": 1.0442695833692566,
+      "learning_rate": 2.924869662871117e-06,
+      "loss": 0.2377,
+      "step": 425
+    },
+    {
+      "epoch": 1.2754491017964071,
+      "grad_norm": 0.9062910588057472,
+      "learning_rate": 2.903498597009136e-06,
+      "loss": 0.2467,
+      "step": 426
+    },
+    {
+      "epoch": 1.278443113772455,
+      "grad_norm": 0.9501269806327622,
+      "learning_rate": 2.8821739015829338e-06,
+      "loss": 0.2475,
+      "step": 427
+    },
+    {
+      "epoch": 1.281437125748503,
+      "grad_norm": 0.8847493600244284,
+      "learning_rate": 2.8608960482523058e-06,
+      "loss": 0.2335,
+      "step": 428
+    },
+    {
+      "epoch": 1.284431137724551,
+      "grad_norm": 0.89853078091105,
+      "learning_rate": 2.839665507640992e-06,
+      "loss": 0.2241,
+      "step": 429
+    },
+    {
+      "epoch": 1.2874251497005988,
+      "grad_norm": 0.8345458778848835,
+      "learning_rate": 2.818482749326272e-06,
+      "loss": 0.2693,
+      "step": 430
+    },
+    {
+      "epoch": 1.2904191616766467,
+      "grad_norm": 0.8373650842866908,
+      "learning_rate": 2.797348241828569e-06,
+      "loss": 0.2321,
+      "step": 431
+    },
+    {
+      "epoch": 1.2934131736526946,
+      "grad_norm": 0.8089985643452108,
+      "learning_rate": 2.776262452601104e-06,
+      "loss": 0.2538,
+      "step": 432
+    },
+    {
+      "epoch": 1.2964071856287425,
+      "grad_norm": 0.929042165229628,
+      "learning_rate": 2.7552258480195348e-06,
+      "loss": 0.2315,
+      "step": 433
+    },
+    {
+      "epoch": 1.2994011976047903,
+      "grad_norm": 0.8300901799503175,
+      "learning_rate": 2.734238893371667e-06,
+      "loss": 0.2249,
+      "step": 434
+    },
+    {
+      "epoch": 1.3023952095808382,
+      "grad_norm": 0.8909248470484058,
+      "learning_rate": 2.7133020528471322e-06,
+      "loss": 0.23,
+      "step": 435
+    },
+    {
+      "epoch": 1.3053892215568863,
+      "grad_norm": 0.835317522371157,
+      "learning_rate": 2.6924157895271563e-06,
+      "loss": 0.2403,
+      "step": 436
+    },
+    {
+      "epoch": 1.3083832335329342,
+      "grad_norm": 0.8647395787653299,
+      "learning_rate": 2.671580565374282e-06,
+      "loss": 0.2601,
+      "step": 437
+    },
+    {
+      "epoch": 1.311377245508982,
+      "grad_norm": 0.928093889607222,
+      "learning_rate": 2.6507968412221763e-06,
+      "loss": 0.2044,
+      "step": 438
+    },
+    {
+      "epoch": 1.31437125748503,
+      "grad_norm": 1.0316703220893109,
+      "learning_rate": 2.6300650767654234e-06,
+      "loss": 0.2279,
+      "step": 439
+    },
+    {
+      "epoch": 1.3173652694610778,
+      "grad_norm": 0.8900386821386082,
+      "learning_rate": 2.6093857305493666e-06,
+      "loss": 0.2345,
+      "step": 440
+    },
+    {
+      "epoch": 1.3203592814371259,
+      "grad_norm": 0.9135592098541239,
+      "learning_rate": 2.588759259959962e-06,
+      "loss": 0.2182,
+      "step": 441
+    },
+    {
+      "epoch": 1.3233532934131738,
+      "grad_norm": 0.8823808480287963,
+      "learning_rate": 2.568186121213658e-06,
+      "loss": 0.2539,
+      "step": 442
+    },
+    {
+      "epoch": 1.3263473053892216,
+      "grad_norm": 0.9188576761856679,
+      "learning_rate": 2.547666769347312e-06,
+      "loss": 0.2455,
+      "step": 443
+    },
+    {
+      "epoch": 1.3293413173652695,
+      "grad_norm": 0.8970132200420725,
+      "learning_rate": 2.5272016582081236e-06,
+      "loss": 0.2945,
+      "step": 444
+    },
+    {
+      "epoch": 1.3323353293413174,
+      "grad_norm": 0.931708535253249,
+      "learning_rate": 2.5067912404435952e-06,
+      "loss": 0.2493,
+      "step": 445
+    },
+    {
+      "epoch": 1.3353293413173652,
+      "grad_norm": 0.913645709906956,
+      "learning_rate": 2.486435967491516e-06,
+      "loss": 0.2289,
+      "step": 446
+    },
+    {
+      "epoch": 1.3383233532934131,
+      "grad_norm": 0.8625374739508064,
+      "learning_rate": 2.4661362895699903e-06,
+      "loss": 0.2193,
+      "step": 447
+    },
+    {
+      "epoch": 1.341317365269461,
+      "grad_norm": 0.9150737245396889,
+      "learning_rate": 2.445892655667462e-06,
+      "loss": 0.299,
+      "step": 448
+    },
+    {
+      "epoch": 1.3443113772455089,
+      "grad_norm": 0.9739809489305545,
+      "learning_rate": 2.425705513532798e-06,
+      "loss": 0.2499,
+      "step": 449
+    },
+    {
+      "epoch": 1.347305389221557,
+      "grad_norm": 0.8383301908743397,
+      "learning_rate": 2.4055753096653795e-06,
+      "loss": 0.2601,
+      "step": 450
+    },
+    {
+      "epoch": 1.3502994011976048,
+      "grad_norm": 0.8327173176755049,
+      "learning_rate": 2.3855024893052286e-06,
+      "loss": 0.2129,
+      "step": 451
+    },
+    {
+      "epoch": 1.3532934131736527,
+      "grad_norm": 0.8647591015863577,
+      "learning_rate": 2.365487496423152e-06,
+      "loss": 0.2297,
+      "step": 452
+    },
+    {
+      "epoch": 1.3562874251497006,
+      "grad_norm": 0.7814006048796464,
+      "learning_rate": 2.3455307737109338e-06,
+      "loss": 0.1944,
+      "step": 453
+    },
+    {
+      "epoch": 1.3592814371257484,
+      "grad_norm": 0.8359729141244253,
+      "learning_rate": 2.3256327625715345e-06,
+      "loss": 0.1983,
+      "step": 454
+    },
+    {
+      "epoch": 1.3622754491017965,
+      "grad_norm": 0.9349340845802615,
+      "learning_rate": 2.3057939031093346e-06,
+      "loss": 0.2217,
+      "step": 455
+    },
+    {
+      "epoch": 1.3652694610778444,
+      "grad_norm": 0.8081200603562876,
+      "learning_rate": 2.2860146341203936e-06,
+      "loss": 0.2141,
+      "step": 456
+    },
+    {
+      "epoch": 1.3682634730538923,
+      "grad_norm": 0.9324708547933994,
+      "learning_rate": 2.2662953930827546e-06,
+      "loss": 0.2882,
+      "step": 457
+    },
+    {
+      "epoch": 1.3712574850299402,
+      "grad_norm": 0.9370627368817129,
+      "learning_rate": 2.2466366161467528e-06,
+      "loss": 0.2324,
+      "step": 458
+    },
+    {
+      "epoch": 1.374251497005988,
+      "grad_norm": 0.9775784913621385,
+      "learning_rate": 2.227038738125385e-06,
+      "loss": 0.2129,
+      "step": 459
+    },
+    {
+      "epoch": 1.377245508982036,
+      "grad_norm": 1.0153366345964563,
+      "learning_rate": 2.207502192484685e-06,
+      "loss": 0.2278,
+      "step": 460
+    },
+    {
+      "epoch": 1.3802395209580838,
+      "grad_norm": 0.8432754551394376,
+      "learning_rate": 2.188027411334131e-06,
+      "loss": 0.2294,
+      "step": 461
+    },
+    {
+      "epoch": 1.3832335329341316,
+      "grad_norm": 0.9453714951431974,
+      "learning_rate": 2.1686148254171012e-06,
+      "loss": 0.2225,
+      "step": 462
+    },
+    {
+      "epoch": 1.3862275449101795,
+      "grad_norm": 1.0293017957323893,
+      "learning_rate": 2.1492648641013305e-06,
+      "loss": 0.2123,
+      "step": 463
+    },
+    {
+      "epoch": 1.3892215568862276,
+      "grad_norm": 0.7973395829106387,
+      "learning_rate": 2.1299779553694323e-06,
+      "loss": 0.2174,
+      "step": 464
+    },
+    {
+      "epoch": 1.3922155688622755,
+      "grad_norm": 0.9822796163667018,
+      "learning_rate": 2.1107545258094135e-06,
+      "loss": 0.226,
+      "step": 465
+    },
+    {
+      "epoch": 1.3952095808383234,
+      "grad_norm": 0.9048202494674366,
+      "learning_rate": 2.0915950006052555e-06,
+      "loss": 0.189,
+      "step": 466
+    },
+    {
+      "epoch": 1.3982035928143712,
+      "grad_norm": 0.8482559459307193,
+      "learning_rate": 2.0724998035274947e-06,
+      "loss": 0.2216,
+      "step": 467
+    },
+    {
+      "epoch": 1.401197604790419,
+      "grad_norm": 0.9006129225448194,
+      "learning_rate": 2.053469356923865e-06,
+      "loss": 0.1872,
+      "step": 468
+    },
+    {
+      "epoch": 1.4041916167664672,
+      "grad_norm": 0.9113520816336781,
+      "learning_rate": 2.0345040817099433e-06,
+      "loss": 0.2328,
+      "step": 469
+    },
+    {
+      "epoch": 1.407185628742515,
+      "grad_norm": 0.7994080695465668,
+      "learning_rate": 2.0156043973598475e-06,
+      "loss": 0.2083,
+      "step": 470
+    },
+    {
+      "epoch": 1.410179640718563,
+      "grad_norm": 0.9704106274414475,
+      "learning_rate": 1.996770721896957e-06,
+      "loss": 0.2377,
+      "step": 471
+    },
+    {
+      "epoch": 1.4131736526946108,
+      "grad_norm": 0.950125781508677,
+      "learning_rate": 1.9780034718846653e-06,
+      "loss": 0.2245,
+      "step": 472
+    },
+    {
+      "epoch": 1.4161676646706587,
+      "grad_norm": 0.8578663798472137,
+      "learning_rate": 1.9593030624171683e-06,
+      "loss": 0.253,
+      "step": 473
+    },
+    {
+      "epoch": 1.4191616766467066,
+      "grad_norm": 0.9856060337714755,
+      "learning_rate": 1.9406699071102774e-06,
+      "loss": 0.2569,
+      "step": 474
+    },
+    {
+      "epoch": 1.4221556886227544,
+      "grad_norm": 0.8365055783466058,
+      "learning_rate": 1.9221044180922833e-06,
+      "loss": 0.2523,
+      "step": 475
+    },
+    {
+      "epoch": 1.4251497005988023,
+      "grad_norm": 0.9166241040139455,
+      "learning_rate": 1.9036070059948253e-06,
+      "loss": 0.2392,
+      "step": 476
+    },
+    {
+      "epoch": 1.4281437125748502,
+      "grad_norm": 1.0362533778177407,
+      "learning_rate": 1.885178079943823e-06,
+      "loss": 0.2679,
+      "step": 477
+    },
+    {
+      "epoch": 1.4311377245508983,
+      "grad_norm": 0.9816009419764054,
+      "learning_rate": 1.866818047550419e-06,
+      "loss": 0.3026,
+      "step": 478
+    },
+    {
+      "epoch": 1.4341317365269461,
+      "grad_norm": 0.8347456795332021,
+      "learning_rate": 1.8485273149019655e-06,
+      "loss": 0.2185,
+      "step": 479
+    },
+    {
+      "epoch": 1.437125748502994,
+      "grad_norm": 0.9091218473568945,
+      "learning_rate": 1.8303062865530407e-06,
+      "loss": 0.2165,
+      "step": 480
+    },
+    {
+      "epoch": 1.4401197604790419,
+      "grad_norm": 0.9154078383701513,
+      "learning_rate": 1.8121553655165058e-06,
+      "loss": 0.2112,
+      "step": 481
+    },
+    {
+      "epoch": 1.4431137724550898,
+      "grad_norm": 0.9032829778128162,
+      "learning_rate": 1.7940749532545832e-06,
+      "loss": 0.265,
+      "step": 482
+    },
+    {
+      "epoch": 1.4461077844311379,
+      "grad_norm": 1.0080697760970814,
+      "learning_rate": 1.7760654496699876e-06,
+      "loss": 0.2227,
+      "step": 483
+    },
+    {
+      "epoch": 1.4491017964071857,
+      "grad_norm": 0.8957733113522632,
+      "learning_rate": 1.7581272530970666e-06,
+      "loss": 0.24,
+      "step": 484
+    },
+    {
+      "epoch": 1.4520958083832336,
+      "grad_norm": 0.8335261496122127,
+      "learning_rate": 1.7402607602930106e-06,
+      "loss": 0.2291,
+      "step": 485
+    },
+    {
+      "epoch": 1.4550898203592815,
+      "grad_norm": 0.9849458838757417,
+      "learning_rate": 1.7224663664290537e-06,
+      "loss": 0.2744,
+      "step": 486
+    },
+    {
+      "epoch": 1.4580838323353293,
+      "grad_norm": 0.8911713531788786,
+      "learning_rate": 1.7047444650817518e-06,
+      "loss": 0.2288,
+      "step": 487
+    },
+    {
+      "epoch": 1.4610778443113772,
+      "grad_norm": 1.0137646108138194,
+      "learning_rate": 1.6870954482242707e-06,
+      "loss": 0.2237,
+      "step": 488
+    },
+    {
+      "epoch": 1.464071856287425,
+      "grad_norm": 0.8871791355821369,
+      "learning_rate": 1.669519706217711e-06,
+      "loss": 0.191,
+      "step": 489
+    },
+    {
+      "epoch": 1.467065868263473,
+      "grad_norm": 0.9401099300088629,
+      "learning_rate": 1.652017627802487e-06,
+      "loss": 0.2461,
+      "step": 490
+    },
+    {
+      "epoch": 1.4700598802395208,
+      "grad_norm": 0.8048016136208538,
+      "learning_rate": 1.6345896000897122e-06,
+      "loss": 0.2149,
+      "step": 491
+    },
+    {
+      "epoch": 1.473053892215569,
+      "grad_norm": 0.8176445572539465,
+      "learning_rate": 1.6172360085526567e-06,
+      "loss": 0.2508,
+      "step": 492
+    },
+    {
+      "epoch": 1.4760479041916168,
+      "grad_norm": 0.8101280568087744,
+      "learning_rate": 1.5999572370182016e-06,
+      "loss": 0.2446,
+      "step": 493
+    },
+    {
+      "epoch": 1.4790419161676647,
+      "grad_norm": 0.9448625335762181,
+      "learning_rate": 1.5827536676583643e-06,
+      "loss": 0.2471,
+      "step": 494
+    },
+    {
+      "epoch": 1.4820359281437125,
+      "grad_norm": 0.8586513112675779,
+      "learning_rate": 1.5656256809818343e-06,
+      "loss": 0.2255,
+      "step": 495
+    },
+    {
+      "epoch": 1.4850299401197604,
+      "grad_norm": 0.9768673466679008,
+      "learning_rate": 1.54857365582557e-06,
+      "loss": 0.1983,
+      "step": 496
+    },
+    {
+      "epoch": 1.4880239520958085,
+      "grad_norm": 0.9006140040891589,
+      "learning_rate": 1.5315979693464039e-06,
+      "loss": 0.24,
+      "step": 497
+    },
+    {
+      "epoch": 1.4910179640718564,
+      "grad_norm": 1.037317719805374,
+      "learning_rate": 1.5146989970127158e-06,
+      "loss": 0.2038,
+      "step": 498
+    },
+    {
+      "epoch": 1.4940119760479043,
+      "grad_norm": 0.9516617265953464,
+      "learning_rate": 1.4978771125961177e-06,
+      "loss": 0.2466,
+      "step": 499
+    },
+    {
+      "epoch": 1.4970059880239521,
+      "grad_norm": 0.81133512200921,
+      "learning_rate": 1.4811326881631937e-06,
+      "loss": 0.2389,
+      "step": 500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8712368300259022,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.2383,
+      "step": 501
+    },
+    {
+      "epoch": 1.5029940119760479,
+      "grad_norm": 0.8912224547341503,
+      "learning_rate": 1.4478776989401949e-06,
+      "loss": 0.2418,
+      "step": 502
+    },
+    {
+      "epoch": 1.5059880239520957,
+      "grad_norm": 0.9081742780158918,
+      "learning_rate": 1.4313678696842559e-06,
+      "loss": 0.2255,
+      "step": 503
+    },
+    {
+      "epoch": 1.5089820359281436,
+      "grad_norm": 0.8912235673005359,
+      "learning_rate": 1.4149369714639856e-06,
+      "loss": 0.22,
+      "step": 504
+    },
+    {
+      "epoch": 1.5119760479041915,
+      "grad_norm": 0.9123351850607487,
+      "learning_rate": 1.3985853676981316e-06,
+      "loss": 0.2372,
+      "step": 505
+    },
+    {
+      "epoch": 1.5149700598802394,
+      "grad_norm": 0.9163696266134105,
+      "learning_rate": 1.3823134200516043e-06,
+      "loss": 0.2237,
+      "step": 506
+    },
+    {
+      "epoch": 1.5179640718562875,
+      "grad_norm": 1.0110038661336658,
+      "learning_rate": 1.366121488427481e-06,
+      "loss": 0.2338,
+      "step": 507
+    },
+    {
+      "epoch": 1.5209580838323353,
+      "grad_norm": 0.9496680513346542,
+      "learning_rate": 1.3500099309590397e-06,
+      "loss": 0.2433,
+      "step": 508
+    },
+    {
+      "epoch": 1.5239520958083832,
+      "grad_norm": 0.9310276996335521,
+      "learning_rate": 1.3339791040018479e-06,
+      "loss": 0.2742,
+      "step": 509
+    },
+    {
+      "epoch": 1.5269461077844313,
+      "grad_norm": 1.0624769483110188,
+      "learning_rate": 1.3180293621258694e-06,
+      "loss": 0.2313,
+      "step": 510
+    },
+    {
+      "epoch": 1.5299401197604792,
+      "grad_norm": 0.9347386553606367,
+      "learning_rate": 1.3021610581076316e-06,
+      "loss": 0.1973,
+      "step": 511
+    },
+    {
+      "epoch": 1.532934131736527,
+      "grad_norm": 1.0200035540914512,
+      "learning_rate": 1.2863745429224145e-06,
+      "loss": 0.2325,
+      "step": 512
+    },
+    {
+      "epoch": 1.535928143712575,
+      "grad_norm": 0.8820492629345995,
+      "learning_rate": 1.270670165736499e-06,
+      "loss": 0.2369,
+      "step": 513
+    },
+    {
+      "epoch": 1.5389221556886228,
+      "grad_norm": 1.0546310430066739,
+      "learning_rate": 1.2550482738994284e-06,
+      "loss": 0.278,
+      "step": 514
+    },
+    {
+      "epoch": 1.5419161676646707,
+      "grad_norm": 0.9206426588636809,
+      "learning_rate": 1.239509212936343e-06,
+      "loss": 0.2285,
+      "step": 515
+    },
+    {
+      "epoch": 1.5449101796407185,
+      "grad_norm": 0.9554122664290174,
+      "learning_rate": 1.22405332654032e-06,
+      "loss": 0.2137,
+      "step": 516
+    },
+    {
+      "epoch": 1.5479041916167664,
+      "grad_norm": 0.7881244236343145,
+      "learning_rate": 1.2086809565647877e-06,
+      "loss": 0.2107,
+      "step": 517
+    },
+    {
+      "epoch": 1.5508982035928143,
+      "grad_norm": 0.9112851247315867,
+      "learning_rate": 1.1933924430159571e-06,
+      "loss": 0.231,
+      "step": 518
+    },
+    {
+      "epoch": 1.5538922155688621,
+      "grad_norm": 0.9191239398127155,
+      "learning_rate": 1.1781881240452958e-06,
+      "loss": 0.2251,
+      "step": 519
+    },
+    {
+      "epoch": 1.55688622754491,
+      "grad_norm": 0.9436054168566522,
+      "learning_rate": 1.1630683359420653e-06,
+      "loss": 0.2413,
+      "step": 520
+    },
+    {
+      "epoch": 1.5598802395209581,
+      "grad_norm": 0.9167771695271543,
+      "learning_rate": 1.1480334131258626e-06,
+      "loss": 0.1854,
+      "step": 521
+    },
+    {
+      "epoch": 1.562874251497006,
+      "grad_norm": 0.986854753217078,
+      "learning_rate": 1.1330836881392405e-06,
+      "loss": 0.2214,
+      "step": 522
+    },
+    {
+      "epoch": 1.5658682634730539,
+      "grad_norm": 0.915423363290195,
+      "learning_rate": 1.11821949164034e-06,
+      "loss": 0.2459,
+      "step": 523
+    },
+    {
+      "epoch": 1.568862275449102,
+      "grad_norm": 1.0260199812872717,
+      "learning_rate": 1.103441152395588e-06,
+      "loss": 0.2862,
+      "step": 524
+    },
+    {
+      "epoch": 1.5718562874251498,
+      "grad_norm": 0.9303227103245472,
+      "learning_rate": 1.088748997272414e-06,
+      "loss": 0.2462,
+      "step": 525
+    },
+    {
+      "epoch": 1.5748502994011977,
+      "grad_norm": 0.9160752971250262,
+      "learning_rate": 1.0741433512320316e-06,
+      "loss": 0.1854,
+      "step": 526
+    },
+    {
+      "epoch": 1.5778443113772456,
+      "grad_norm": 1.0599240939257424,
+      "learning_rate": 1.0596245373222424e-06,
+      "loss": 0.2206,
+      "step": 527
+    },
+    {
+      "epoch": 1.5808383233532934,
+      "grad_norm": 0.899009303584087,
+      "learning_rate": 1.045192876670298e-06,
+      "loss": 0.253,
+      "step": 528
+    },
+    {
+      "epoch": 1.5838323353293413,
+      "grad_norm": 0.8727433242418636,
+      "learning_rate": 1.0308486884757868e-06,
+      "loss": 0.261,
+      "step": 529
+    },
+    {
+      "epoch": 1.5868263473053892,
+      "grad_norm": 0.8286399656871455,
+      "learning_rate": 1.0165922900035886e-06,
+      "loss": 0.2353,
+      "step": 530
+    },
+    {
+      "epoch": 1.589820359281437,
+      "grad_norm": 0.8764735752385893,
+      "learning_rate": 1.0024239965768417e-06,
+      "loss": 0.2229,
+      "step": 531
+    },
+    {
+      "epoch": 1.592814371257485,
+      "grad_norm": 0.9219440935456223,
+      "learning_rate": 9.883441215699824e-07,
+      "loss": 0.1751,
+      "step": 532
+    },
+    {
+      "epoch": 1.5958083832335328,
+      "grad_norm": 1.080747710783597,
+      "learning_rate": 9.74352976401805e-07,
+      "loss": 0.2792,
+      "step": 533
+    },
+    {
+      "epoch": 1.5988023952095807,
+      "grad_norm": 1.0084037044161969,
+      "learning_rate": 9.604508705285765e-07,
+      "loss": 0.2342,
+      "step": 534
+    },
+    {
+      "epoch": 1.6017964071856288,
+      "grad_norm": 0.9488711476032898,
+      "learning_rate": 9.466381114371942e-07,
+      "loss": 0.2358,
+      "step": 535
+    },
+    {
+      "epoch": 1.6047904191616766,
+      "grad_norm": 0.9213491807397138,
+      "learning_rate": 9.329150046383773e-07,
+      "loss": 0.2299,
+      "step": 536
+    },
+    {
+      "epoch": 1.6077844311377245,
+      "grad_norm": 0.9871797150588095,
+      "learning_rate": 9.192818536599213e-07,
+      "loss": 0.243,
+      "step": 537
+    },
+    {
+      "epoch": 1.6107784431137726,
+      "grad_norm": 0.8941476823598501,
+      "learning_rate": 9.057389600399719e-07,
+      "loss": 0.243,
+      "step": 538
+    },
+    {
+      "epoch": 1.6137724550898205,
+      "grad_norm": 0.8562245477543875,
+      "learning_rate": 8.922866233203681e-07,
+      "loss": 0.2187,
+      "step": 539
+    },
+    {
+      "epoch": 1.6167664670658684,
+      "grad_norm": 0.7452803883355023,
+      "learning_rate": 8.789251410400024e-07,
+      "loss": 0.1749,
+      "step": 540
+    },
+    {
+      "epoch": 1.6197604790419162,
+      "grad_norm": 0.7355325160042411,
+      "learning_rate": 8.65654808728259e-07,
+      "loss": 0.2249,
+      "step": 541
+    },
+    {
+      "epoch": 1.622754491017964,
+      "grad_norm": 1.0984577194709262,
+      "learning_rate": 8.524759198984567e-07,
+      "loss": 0.2139,
+      "step": 542
+    },
+    {
+      "epoch": 1.625748502994012,
+      "grad_norm": 0.896782748741311,
+      "learning_rate": 8.393887660413719e-07,
+      "loss": 0.2276,
+      "step": 543
+    },
+    {
+      "epoch": 1.6287425149700598,
+      "grad_norm": 1.075942766285842,
+      "learning_rate": 8.263936366187825e-07,
+      "loss": 0.2716,
+      "step": 544
+    },
+    {
+      "epoch": 1.6317365269461077,
+      "grad_norm": 0.8861486842740338,
+      "learning_rate": 8.134908190570723e-07,
+      "loss": 0.2502,
+      "step": 545
+    },
+    {
+      "epoch": 1.6347305389221556,
+      "grad_norm": 0.7935283743028442,
+      "learning_rate": 8.006805987408705e-07,
+      "loss": 0.1715,
+      "step": 546
+    },
+    {
+      "epoch": 1.6377245508982035,
+      "grad_norm": 0.9590256087574384,
+      "learning_rate": 7.879632590067354e-07,
+      "loss": 0.2354,
+      "step": 547
+    },
+    {
+      "epoch": 1.6407185628742516,
+      "grad_norm": 0.9975426487033544,
+      "learning_rate": 7.753390811368972e-07,
+      "loss": 0.2635,
+      "step": 548
+    },
+    {
+      "epoch": 1.6437125748502994,
+      "grad_norm": 0.9145620885422897,
+      "learning_rate": 7.628083443530287e-07,
+      "loss": 0.2333,
+      "step": 549
+    },
+    {
+      "epoch": 1.6467065868263473,
+      "grad_norm": 1.0373271576164917,
+      "learning_rate": 7.503713258100726e-07,
+      "loss": 0.2545,
+      "step": 550
+    },
+    {
+      "epoch": 1.6497005988023952,
+      "grad_norm": 0.8593826234585763,
+      "learning_rate": 7.380283005901084e-07,
+      "loss": 0.2391,
+      "step": 551
+    },
+    {
+      "epoch": 1.6526946107784433,
+      "grad_norm": 0.8320133036949128,
+      "learning_rate": 7.257795416962754e-07,
+      "loss": 0.2127,
+      "step": 552
+    },
+    {
+      "epoch": 1.6556886227544911,
+      "grad_norm": 0.8847180137824004,
+      "learning_rate": 7.136253200467231e-07,
+      "loss": 0.2297,
+      "step": 553
+    },
+    {
+      "epoch": 1.658682634730539,
+      "grad_norm": 0.9185850186489214,
+      "learning_rate": 7.015659044686307e-07,
+      "loss": 0.2274,
+      "step": 554
+    },
+    {
+      "epoch": 1.6616766467065869,
+      "grad_norm": 0.9169111137668451,
+      "learning_rate": 6.896015616922535e-07,
+      "loss": 0.2291,
+      "step": 555
+    },
+    {
+      "epoch": 1.6646706586826348,
+      "grad_norm": 0.856353739728233,
+      "learning_rate": 6.777325563450282e-07,
+      "loss": 0.2679,
+      "step": 556
+    },
+    {
+      "epoch": 1.6676646706586826,
+      "grad_norm": 0.9043875388097713,
+      "learning_rate": 6.659591509457125e-07,
+      "loss": 0.2271,
+      "step": 557
+    },
+    {
+      "epoch": 1.6706586826347305,
+      "grad_norm": 0.9650530076188746,
+      "learning_rate": 6.542816058985896e-07,
+      "loss": 0.2354,
+      "step": 558
+    },
+    {
+      "epoch": 1.6736526946107784,
+      "grad_norm": 0.954636182956266,
+      "learning_rate": 6.427001794876974e-07,
+      "loss": 0.2277,
+      "step": 559
+    },
+    {
+      "epoch": 1.6766467065868262,
+      "grad_norm": 0.8374283590322238,
+      "learning_rate": 6.312151278711237e-07,
+      "loss": 0.2476,
+      "step": 560
+    },
+    {
+      "epoch": 1.6796407185628741,
+      "grad_norm": 0.8399538250559132,
+      "learning_rate": 6.198267050753387e-07,
+      "loss": 0.2284,
+      "step": 561
+    },
+    {
+      "epoch": 1.6826347305389222,
+      "grad_norm": 0.9345900962466387,
+      "learning_rate": 6.085351629895736e-07,
+      "loss": 0.2277,
+      "step": 562
+    },
+    {
+      "epoch": 1.68562874251497,
+      "grad_norm": 0.9966297559348759,
+      "learning_rate": 5.973407513602514e-07,
+      "loss": 0.2054,
+      "step": 563
+    },
+    {
+      "epoch": 1.688622754491018,
+      "grad_norm": 0.9308977097223865,
+      "learning_rate": 5.862437177854629e-07,
+      "loss": 0.249,
+      "step": 564
+    },
+    {
+      "epoch": 1.6916167664670658,
+      "grad_norm": 1.0515595335934342,
+      "learning_rate": 5.752443077094927e-07,
+      "loss": 0.2098,
+      "step": 565
+    },
+    {
+      "epoch": 1.694610778443114,
+      "grad_norm": 0.9038962722445684,
+      "learning_rate": 5.643427644173838e-07,
+      "loss": 0.2294,
+      "step": 566
+    },
+    {
+      "epoch": 1.6976047904191618,
+      "grad_norm": 0.8065933884675035,
+      "learning_rate": 5.535393290295643e-07,
+      "loss": 0.2081,
+      "step": 567
+    },
+    {
+      "epoch": 1.7005988023952097,
+      "grad_norm": 0.9019474894135322,
+      "learning_rate": 5.428342404965076e-07,
+      "loss": 0.2085,
+      "step": 568
+    },
+    {
+      "epoch": 1.7035928143712575,
+      "grad_norm": 0.918436592444138,
+      "learning_rate": 5.322277355934557e-07,
+      "loss": 0.2602,
+      "step": 569
+    },
+    {
+      "epoch": 1.7065868263473054,
+      "grad_norm": 0.8691317529476447,
+      "learning_rate": 5.217200489151714e-07,
+      "loss": 0.2372,
+      "step": 570
+    },
+    {
+      "epoch": 1.7095808383233533,
+      "grad_norm": 0.8343949227696129,
+      "learning_rate": 5.113114128707592e-07,
+      "loss": 0.1918,
+      "step": 571
+    },
+    {
+      "epoch": 1.7125748502994012,
+      "grad_norm": 0.953652304613386,
+      "learning_rate": 5.010020576785174e-07,
+      "loss": 0.2491,
+      "step": 572
+    },
+    {
+      "epoch": 1.715568862275449,
+      "grad_norm": 0.8124988839406792,
+      "learning_rate": 4.907922113608532e-07,
+      "loss": 0.1852,
+      "step": 573
+    },
+    {
+      "epoch": 1.718562874251497,
+      "grad_norm": 0.922773122795006,
+      "learning_rate": 4.806820997392325e-07,
+      "loss": 0.2289,
+      "step": 574
+    },
+    {
+      "epoch": 1.7215568862275448,
+      "grad_norm": 0.8340937152958176,
+      "learning_rate": 4.7067194642919036e-07,
+      "loss": 0.2223,
+      "step": 575
+    },
+    {
+      "epoch": 1.7245508982035929,
+      "grad_norm": 0.95537765114133,
+      "learning_rate": 4.607619728353818e-07,
+      "loss": 0.2138,
+      "step": 576
+    },
+    {
+      "epoch": 1.7275449101796407,
+      "grad_norm": 0.8369921763931705,
+      "learning_rate": 4.50952398146689e-07,
+      "loss": 0.2271,
+      "step": 577
+    },
+    {
+      "epoch": 1.7305389221556886,
+      "grad_norm": 0.8468425311718798,
+      "learning_rate": 4.4124343933136525e-07,
+      "loss": 0.2205,
+      "step": 578
+    },
+    {
+      "epoch": 1.7335329341317365,
+      "grad_norm": 0.8289388386941247,
+      "learning_rate": 4.3163531113224466e-07,
+      "loss": 0.1981,
+      "step": 579
+    },
+    {
+      "epoch": 1.7365269461077846,
+      "grad_norm": 0.9146570918536742,
+      "learning_rate": 4.221282260619891e-07,
+      "loss": 0.2398,
+      "step": 580
+    },
+    {
+      "epoch": 1.7395209580838324,
+      "grad_norm": 0.8297604996586672,
+      "learning_rate": 4.127223943983849e-07,
+      "loss": 0.2154,
+      "step": 581
+    },
+    {
+      "epoch": 1.7425149700598803,
+      "grad_norm": 0.837589147748755,
+      "learning_rate": 4.03418024179697e-07,
+      "loss": 0.2152,
+      "step": 582
+    },
+    {
+      "epoch": 1.7455089820359282,
+      "grad_norm": 0.8644075202235623,
+      "learning_rate": 3.9421532120006544e-07,
+      "loss": 0.2036,
+      "step": 583
+    },
+    {
+      "epoch": 1.748502994011976,
+      "grad_norm": 0.8659250328332493,
+      "learning_rate": 3.851144890049535e-07,
+      "loss": 0.2218,
+      "step": 584
+    },
+    {
+      "epoch": 1.751497005988024,
+      "grad_norm": 0.9583473617849272,
+      "learning_rate": 3.761157288866418e-07,
+      "loss": 0.279,
+      "step": 585
+    },
+    {
+      "epoch": 1.7544910179640718,
+      "grad_norm": 0.9245215526376561,
+      "learning_rate": 3.672192398797858e-07,
+      "loss": 0.2244,
+      "step": 586
+    },
+    {
+      "epoch": 1.7574850299401197,
+      "grad_norm": 0.9649161709724987,
+      "learning_rate": 3.58425218757002e-07,
+      "loss": 0.22,
+      "step": 587
+    },
+    {
+      "epoch": 1.7604790419161676,
+      "grad_norm": 0.894173288282072,
+      "learning_rate": 3.497338600245254e-07,
+      "loss": 0.2162,
+      "step": 588
+    },
+    {
+      "epoch": 1.7634730538922154,
+      "grad_norm": 0.9065861440899793,
+      "learning_rate": 3.4114535591790233e-07,
+      "loss": 0.2059,
+      "step": 589
+    },
+    {
+      "epoch": 1.7664670658682635,
+      "grad_norm": 0.9686267954709409,
+      "learning_rate": 3.326598963977395e-07,
+      "loss": 0.2085,
+      "step": 590
+    },
+    {
+      "epoch": 1.7694610778443114,
+      "grad_norm": 1.0006343137554319,
+      "learning_rate": 3.242776691455013e-07,
+      "loss": 0.2438,
+      "step": 591
+    },
+    {
+      "epoch": 1.7724550898203593,
+      "grad_norm": 0.7332242297499415,
+      "learning_rate": 3.159988595593616e-07,
+      "loss": 0.1991,
+      "step": 592
+    },
+    {
+      "epoch": 1.7754491017964071,
+      "grad_norm": 0.8878633844075399,
+      "learning_rate": 3.078236507501015e-07,
+      "loss": 0.1988,
+      "step": 593
+    },
+    {
+      "epoch": 1.7784431137724552,
+      "grad_norm": 0.8631090766938814,
+      "learning_rate": 2.9975222353705757e-07,
+      "loss": 0.2387,
+      "step": 594
+    },
+    {
+      "epoch": 1.781437125748503,
+      "grad_norm": 0.8850786727974973,
+      "learning_rate": 2.917847564441256e-07,
+      "loss": 0.2561,
+      "step": 595
+    },
+    {
+      "epoch": 1.784431137724551,
+      "grad_norm": 0.9310837214986993,
+      "learning_rate": 2.839214256958106e-07,
+      "loss": 0.2194,
+      "step": 596
+    },
+    {
+      "epoch": 1.7874251497005988,
+      "grad_norm": 0.9377101843539455,
+      "learning_rate": 2.7616240521332884e-07,
+      "loss": 0.217,
+      "step": 597
+    },
+    {
+      "epoch": 1.7904191616766467,
+      "grad_norm": 0.8239091250636275,
+      "learning_rate": 2.6850786661076047e-07,
+      "loss": 0.2331,
+      "step": 598
+    },
+    {
+      "epoch": 1.7934131736526946,
+      "grad_norm": 0.9118979148964735,
+      "learning_rate": 2.6095797919125533e-07,
+      "loss": 0.2103,
+      "step": 599
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "grad_norm": 0.8534667314237735,
+      "learning_rate": 2.5351290994328703e-07,
+      "loss": 0.2395,
+      "step": 600
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "eval_loss": 0.32940974831581116,
+      "eval_runtime": 3.5774,
+      "eval_samples_per_second": 15.095,
+      "eval_steps_per_second": 3.913,
+      "step": 600
+    },
+    {
+      "epoch": 1.7994011976047903,
+      "grad_norm": 0.8119494414606993,
+      "learning_rate": 2.4617282353696093e-07,
+      "loss": 0.2028,
+      "step": 601
+    },
+    {
+      "epoch": 1.8023952095808382,
+      "grad_norm": 0.9678902123635491,
+      "learning_rate": 2.3893788232036807e-07,
+      "loss": 0.2794,
+      "step": 602
+    },
+    {
+      "epoch": 1.805389221556886,
+      "grad_norm": 0.9184998292750141,
+      "learning_rate": 2.318082463160032e-07,
+      "loss": 0.2605,
+      "step": 603
+    },
+    {
+      "epoch": 1.8083832335329342,
+      "grad_norm": 0.8258050559975173,
+      "learning_rate": 2.2478407321721295e-07,
+      "loss": 0.2173,
+      "step": 604
+    },
+    {
+      "epoch": 1.811377245508982,
+      "grad_norm": 0.8162702386622641,
+      "learning_rate": 2.1786551838471892e-07,
+      "loss": 0.2233,
+      "step": 605
+    },
+    {
+      "epoch": 1.81437125748503,
+      "grad_norm": 0.995800718266279,
+      "learning_rate": 2.1105273484317402e-07,
+      "loss": 0.1913,
+      "step": 606
+    },
+    {
+      "epoch": 1.8173652694610778,
+      "grad_norm": 0.8339609053360876,
+      "learning_rate": 2.043458732777831e-07,
+      "loss": 0.1993,
+      "step": 607
+    },
+    {
+      "epoch": 1.8203592814371259,
+      "grad_norm": 0.9485170803779748,
+      "learning_rate": 1.9774508203096843e-07,
+      "loss": 0.2336,
+      "step": 608
+    },
+    {
+      "epoch": 1.8233532934131738,
+      "grad_norm": 0.8176409009358556,
+      "learning_rate": 1.9125050709908388e-07,
+      "loss": 0.1949,
+      "step": 609
+    },
+    {
+      "epoch": 1.8263473053892216,
+      "grad_norm": 0.9188908452030895,
+      "learning_rate": 1.8486229212919482e-07,
+      "loss": 0.2364,
+      "step": 610
+    },
+    {
+      "epoch": 1.8293413173652695,
+      "grad_norm": 0.8255676677137985,
+      "learning_rate": 1.7858057841589281e-07,
+      "loss": 0.2234,
+      "step": 611
+    },
+    {
+      "epoch": 1.8323353293413174,
+      "grad_norm": 1.0315162352494296,
+      "learning_rate": 1.7240550489817652e-07,
+      "loss": 0.2477,
+      "step": 612
+    },
+    {
+      "epoch": 1.8353293413173652,
+      "grad_norm": 1.0211294700959108,
+      "learning_rate": 1.66337208156373e-07,
+      "loss": 0.2168,
+      "step": 613
+    },
+    {
+      "epoch": 1.8383233532934131,
+      "grad_norm": 0.9026794164390517,
+      "learning_rate": 1.6037582240912175e-07,
+      "loss": 0.2368,
+      "step": 614
+    },
+    {
+      "epoch": 1.841317365269461,
+      "grad_norm": 0.8776064474161454,
+      "learning_rate": 1.5452147951040165e-07,
+      "loss": 0.2323,
+      "step": 615
+    },
+    {
+      "epoch": 1.8443113772455089,
+      "grad_norm": 1.009690999308888,
+      "learning_rate": 1.4877430894662037e-07,
+      "loss": 0.2649,
+      "step": 616
+    },
+    {
+      "epoch": 1.8473053892215567,
+      "grad_norm": 0.8955062054578431,
+      "learning_rate": 1.4313443783374405e-07,
+      "loss": 0.21,
+      "step": 617
+    },
+    {
+      "epoch": 1.8502994011976048,
+      "grad_norm": 0.8223285759016997,
+      "learning_rate": 1.3760199091449045e-07,
+      "loss": 0.1883,
+      "step": 618
+    },
+    {
+      "epoch": 1.8532934131736527,
+      "grad_norm": 0.8040628257364225,
+      "learning_rate": 1.3217709055556638e-07,
+      "loss": 0.2,
+      "step": 619
+    },
+    {
+      "epoch": 1.8562874251497006,
+      "grad_norm": 0.858235644570802,
+      "learning_rate": 1.268598567449647e-07,
+      "loss": 0.2267,
+      "step": 620
+    },
+    {
+      "epoch": 1.8592814371257484,
+      "grad_norm": 0.9641435650840584,
+      "learning_rate": 1.2165040708930763e-07,
+      "loss": 0.2226,
+      "step": 621
+    },
+    {
+      "epoch": 1.8622754491017965,
+      "grad_norm": 0.7799819481928281,
+      "learning_rate": 1.1654885681124661e-07,
+      "loss": 0.221,
+      "step": 622
+    },
+    {
+      "epoch": 1.8652694610778444,
+      "grad_norm": 0.9100257311468898,
+      "learning_rate": 1.1155531874691372e-07,
+      "loss": 0.2527,
+      "step": 623
+    },
+    {
+      "epoch": 1.8682634730538923,
+      "grad_norm": 0.8770022413908396,
+      "learning_rate": 1.0666990334342708e-07,
+      "loss": 0.2032,
+      "step": 624
+    },
+    {
+      "epoch": 1.8712574850299402,
+      "grad_norm": 0.8674898686200286,
+      "learning_rate": 1.0189271865644445e-07,
+      "loss": 0.228,
+      "step": 625
+    },
+    {
+      "epoch": 1.874251497005988,
+      "grad_norm": 1.1343808792109478,
+      "learning_rate": 9.722387034777847e-08,
+      "loss": 0.2698,
+      "step": 626
+    },
+    {
+      "epoch": 1.877245508982036,
+      "grad_norm": 0.9294444437721019,
+      "learning_rate": 9.266346168305518e-08,
+      "loss": 0.2618,
+      "step": 627
+    },
+    {
+      "epoch": 1.8802395209580838,
+      "grad_norm": 0.8206454566517795,
+      "learning_rate": 8.821159352943142e-08,
+      "loss": 0.2211,
+      "step": 628
+    },
+    {
+      "epoch": 1.8832335329341316,
+      "grad_norm": 0.9086609286270629,
+      "learning_rate": 8.38683643533661e-08,
+      "loss": 0.2339,
+      "step": 629
+    },
+    {
+      "epoch": 1.8862275449101795,
+      "grad_norm": 0.9799968800296519,
+      "learning_rate": 7.963387021843683e-08,
+      "loss": 0.2335,
+      "step": 630
+    },
+    {
+      "epoch": 1.8892215568862274,
+      "grad_norm": 0.928853130443077,
+      "learning_rate": 7.550820478322285e-08,
+      "loss": 0.2521,
+      "step": 631
+    },
+    {
+      "epoch": 1.8922155688622755,
+      "grad_norm": 0.8598786368119536,
+      "learning_rate": 7.149145929922607e-08,
+      "loss": 0.1957,
+      "step": 632
+    },
+    {
+      "epoch": 1.8952095808383234,
+      "grad_norm": 0.915068110335021,
+      "learning_rate": 6.758372260885714e-08,
+      "loss": 0.2442,
+      "step": 633
+    },
+    {
+      "epoch": 1.8982035928143712,
+      "grad_norm": 1.0142945468007212,
+      "learning_rate": 6.378508114346982e-08,
+      "loss": 0.2215,
+      "step": 634
+    },
+    {
+      "epoch": 1.9011976047904193,
+      "grad_norm": 1.0404226389663245,
+      "learning_rate": 6.009561892144744e-08,
+      "loss": 0.2193,
+      "step": 635
+    },
+    {
+      "epoch": 1.9041916167664672,
+      "grad_norm": 0.9059426399430126,
+      "learning_rate": 5.651541754634726e-08,
+      "loss": 0.204,
+      "step": 636
+    },
+    {
+      "epoch": 1.907185628742515,
+      "grad_norm": 0.9607271314104048,
+      "learning_rate": 5.304455620509297e-08,
+      "loss": 0.2397,
+      "step": 637
+    },
+    {
+      "epoch": 1.910179640718563,
+      "grad_norm": 0.8903471611935083,
+      "learning_rate": 4.968311166622553e-08,
+      "loss": 0.2259,
+      "step": 638
+    },
+    {
+      "epoch": 1.9131736526946108,
+      "grad_norm": 0.9739038012140303,
+      "learning_rate": 4.643115827820399e-08,
+      "loss": 0.2012,
+      "step": 639
+    },
+    {
+      "epoch": 1.9161676646706587,
+      "grad_norm": 0.9671923210116596,
+      "learning_rate": 4.328876796776071e-08,
+      "loss": 0.2611,
+      "step": 640
+    },
+    {
+      "epoch": 1.9191616766467066,
+      "grad_norm": 0.9810030299257178,
+      "learning_rate": 4.0256010238310936e-08,
+      "loss": 0.2237,
+      "step": 641
+    },
+    {
+      "epoch": 1.9221556886227544,
+      "grad_norm": 0.859877438692762,
+      "learning_rate": 3.733295216841626e-08,
+      "loss": 0.2419,
+      "step": 642
+    },
+    {
+      "epoch": 1.9251497005988023,
+      "grad_norm": 0.9007188955813349,
+      "learning_rate": 3.451965841029914e-08,
+      "loss": 0.2246,
+      "step": 643
+    },
+    {
+      "epoch": 1.9281437125748502,
+      "grad_norm": 0.8394254102020391,
+      "learning_rate": 3.181619118841517e-08,
+      "loss": 0.2666,
+      "step": 644
+    },
+    {
+      "epoch": 1.931137724550898,
+      "grad_norm": 1.0110764984923726,
+      "learning_rate": 2.9222610298074717e-08,
+      "loss": 0.2124,
+      "step": 645
+    },
+    {
+      "epoch": 1.9341317365269461,
+      "grad_norm": 0.9137239418802343,
+      "learning_rate": 2.673897310412288e-08,
+      "loss": 0.2583,
+      "step": 646
+    },
+    {
+      "epoch": 1.937125748502994,
+      "grad_norm": 0.7529780732128938,
+      "learning_rate": 2.4365334539667717e-08,
+      "loss": 0.1931,
+      "step": 647
+    },
+    {
+      "epoch": 1.9401197604790419,
+      "grad_norm": 0.8730933596034636,
+      "learning_rate": 2.210174710486679e-08,
+      "loss": 0.2138,
+      "step": 648
+    },
+    {
+      "epoch": 1.94311377245509,
+      "grad_norm": 0.8365012170520347,
+      "learning_rate": 1.99482608657664e-08,
+      "loss": 0.211,
+      "step": 649
+    },
+    {
+      "epoch": 1.9461077844311379,
+      "grad_norm": 0.9272064842770233,
+      "learning_rate": 1.7904923453193056e-08,
+      "loss": 0.2508,
+      "step": 650
+    },
+    {
+      "epoch": 1.9491017964071857,
+      "grad_norm": 0.8486071913399128,
+      "learning_rate": 1.5971780061701524e-08,
+      "loss": 0.2129,
+      "step": 651
+    },
+    {
+      "epoch": 1.9520958083832336,
+      "grad_norm": 0.9006672267111283,
+      "learning_rate": 1.4148873448573408e-08,
+      "loss": 0.2009,
+      "step": 652
+    },
+    {
+      "epoch": 1.9550898203592815,
+      "grad_norm": 0.8278135537668184,
+      "learning_rate": 1.2436243932872349e-08,
+      "loss": 0.2337,
+      "step": 653
+    },
+    {
+      "epoch": 1.9580838323353293,
+      "grad_norm": 0.8876450837281434,
+      "learning_rate": 1.0833929394552523e-08,
+      "loss": 0.1976,
+      "step": 654
+    },
+    {
+      "epoch": 1.9610778443113772,
+      "grad_norm": 0.8423533893800599,
+      "learning_rate": 9.341965273621522e-09,
+      "loss": 0.2421,
+      "step": 655
+    },
+    {
+      "epoch": 1.964071856287425,
+      "grad_norm": 0.9317422471032494,
+      "learning_rate": 7.96038456935322e-09,
+      "loss": 0.2387,
+      "step": 656
+    },
+    {
+      "epoch": 1.967065868263473,
+      "grad_norm": 0.916442743765092,
+      "learning_rate": 6.6892178395611125e-09,
+      "loss": 0.2567,
+      "step": 657
+    },
+    {
+      "epoch": 1.9700598802395208,
+      "grad_norm": 1.1599640798849271,
+      "learning_rate": 5.528493199922769e-09,
+      "loss": 0.2543,
+      "step": 658
+    },
+    {
+      "epoch": 1.9730538922155687,
+      "grad_norm": 0.8906115635398465,
+      "learning_rate": 4.478236323355312e-09,
+      "loss": 0.2055,
+      "step": 659
+    },
+    {
+      "epoch": 1.9760479041916168,
+      "grad_norm": 0.986996432255188,
+      "learning_rate": 3.538470439448105e-09,
+      "loss": 0.2812,
+      "step": 660
+    },
+    {
+      "epoch": 1.9790419161676647,
+      "grad_norm": 0.8660770888742138,
+      "learning_rate": 2.709216333952602e-09,
+      "loss": 0.2395,
+      "step": 661
+    },
+    {
+      "epoch": 1.9820359281437125,
+      "grad_norm": 0.8200642556953853,
+      "learning_rate": 1.9904923483171632e-09,
+      "loss": 0.1981,
+      "step": 662
+    },
+    {
+      "epoch": 1.9850299401197606,
+      "grad_norm": 0.9519722214351811,
+      "learning_rate": 1.3823143792851545e-09,
+      "loss": 0.2791,
+      "step": 663
+    },
+    {
+      "epoch": 1.9880239520958085,
+      "grad_norm": 1.0220351222216668,
+      "learning_rate": 8.846958785418969e-10,
+      "loss": 0.2629,
+      "step": 664
+    },
+    {
+      "epoch": 1.9910179640718564,
+      "grad_norm": 0.896429038531988,
+      "learning_rate": 4.97647852417682e-10,
+      "loss": 0.2034,
+      "step": 665
+    },
+    {
+      "epoch": 1.9940119760479043,
+      "grad_norm": 0.8392074572982221,
+      "learning_rate": 2.2117886164407797e-10,
+      "loss": 0.2096,
+      "step": 666
+    },
+    {
+      "epoch": 1.9970059880239521,
+      "grad_norm": 0.7640885896999279,
+      "learning_rate": 5.529502116519148e-11,
+      "loss": 0.1963,
+      "step": 667
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8428548981366851,
+      "learning_rate": 0.0,
+      "loss": 0.1478,
+      "step": 668
+    },
+    {
+      "epoch": 2.0,
+      "step": 668,
+      "total_flos": 35357291446272.0,
+      "train_loss": 0.2912723551877958,
+      "train_runtime": 1590.1983,
+      "train_samples_per_second": 6.702,
+      "train_steps_per_second": 0.42
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 668,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 35357291446272.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}